The opinion delivered on 12 June 2025 by Advocate General Maciej Szpunar (“AG“) before the European Court of Justice (“ECJ“) in Case C-286/24 Meliá Hotels International v Associação Ius Omnibus has a clarifying effect in the still relatively new regime of private enforcement of cartel law with regard to the handling of trade secrets, in particular regarding the requirements for the disclosure of evidence prior to the initiation of legal action for damages, and the related challenges with regard to the disclosure and protection of trade secrets.

II. Facts of the case

By decision of 21 February 2020, the European Commission found that Meliá Hotels International had infringed Article 101 TFEU and Article 53 of the EEA Agreement. Meliá had treated consumers differently on the basis of their nationality or country of residence, resulting in a restriction on active and passive sales of hotel accommodation. As a result of their co-operation, the fine imposed on Meliá was reduced.

Associação Ius Omnibus, a Portuguese consumer rights organisation, filed a special action for disclosure of documents in Meliá’s possession. These documents were deemed necessary to determine and prove the extent and effects of the anti-competitive behaviour and the harm caused to consumers. This disclosure was to precede a possible collective action for damages.

The court of first instance ruled in favour of Ius Omnibus, and the court of appeal upheld this ruling (para. 67). Subsequently, Meliá lodged an extraordinary appeal with the Supremo Tribunal de Justiça (Supreme Court, Portugal), which resulted in the proceedings being stayed and the referral of the following questions to the ECJ for a preliminary ruling:

1. Does Article 5(1) of Directive 2014/104 apply to an action for access to evidence prior to an action for damages within the meaning of Article 2(4) of that directive?

If the answer to that question is in the affirmative:

2. Does the requirement of plausibility of harm arising from Article 5(1) of Directive 2014/104 always require the applicant to show that, in the particular case, harm to the consumers represented – in this case consumers resident in Portugal – is more likely than the opposite?[1]
3. Can the national courts base the criterion of the plausibility of the harm under Article 5(1) of Directive 2014/104 solely on the existence of a decision of the competent competition authorities? In particular, how does the fact that it is a decision in settlement proceedings concerning an intended vertical infringement of European competition law affect this assessment?

III. Key points of the opinion in relation to the disclosure of trade secrets

1. The temporal applicability of the disclosure rules

The AG’s opinion suggests that even though Directive 2014/104/EU primarily addresses requests for disclosure of evidence within the context of a damages action, such a request made procedurally before the damages action is initiated may still fall under the directive’s scope (paras. 22, 28). In certain scenarios, it can be assumed that such an application is made in the context of a damages action or contingent upon the initiation of such an action. This is typically the case when a damages action must be filed in anticipation of sanctions, either shortly after the submission of a disclosure request, where the credibility of the damages claim is assessed, or within a brief period following the approval of the request.

2. Costs and scope of the disclosure claim as decisive criteria of proportionality

The proportionality assessment began with an examination of whether Associação Ius Omnibus had sufficiently demonstrated the plausibility of its claim for damages. The AG emphasised that the disclosure of evidence must be proportionate. This means that the national courts must weigh up whether disclosure is proportionate to the legitimate interests of the parties. In particular, the scope and costs of disclosure and the avoidance of an untargeted search for information must be taken into account. If the evidence to be disclosed contains confidential information, appropriate protective measures must be taken to protect this information (para. 3, 35).

3. Disclosure of evidence and trade secrets subject to protective measures

A key issue in the proceedings was establishing the conditions under which trade secrets can be protected during the disclosure of evidence. The AG emphasized that national courts must consider the legitimate interests of all involved parties and third parties when ordering evidence disclosure. It is crucial for courts to assess whether the evidence contains confidential information and to ensure appropriate precautions are in place to protect this information (paras. 3, 35). Consequently, national courts are required to ensure that such information is disclosed only to the extent necessary for adjudicating the case. The AG outlined that national courts have the responsibility to protect business secrets through two main options: (i) implementing confidentiality measures or (ii) restricting access to certain information. However, the AG did not provide a detailed evaluation of these specific measures or their precise nature in a comparative context. The European Commission’s Communication on the protection of confidential information by national courts in private enforcement of EU competition law, dated 22 July 2020, offers further guidance on the envisaged confidentiality measures.

4. Plausibility of the claim for damages

While an order from authority finding an infringement of competition law relieves the court dealing with an application for disclosure of evidence of the obligation to examine whether the infringement is plausible in the light of the factual circumstances and the available evidence (para. 45), the unlawfulness of the alleged behaviour is only one of the prerequisites for liability for an infringement of competition law.

With regards to the question of the degree of plausibility of the damages claims (question 3), the AG stated that a decision finding an infringement of competition law itself is not sufficient to establish the plausibility of a claim for damages. Additionally, the fact that this decision concerns a vertical restraint by object and was issued in the context of settlement proceedings does not alter this assessment (para. 62). Instead, the onus is on the plaintiff to provide sufficient evidence to support their claim in a plausible manner, using facts and evidence that are reasonably accessible (paras. 36, 43). The AG notes that this criterion should therefore also be able to be fulfilled in the case of incomplete information (para. 84). In any case, it is imperative to demonstrate that the conditions giving rise to liability are more likely to be fulfilled than the opposite (para. 87).[2]

Finally, in the context of the application for disclosure of evidence under Article 5(1) of Directive 2014/104/EU, the requisite degree of plausibility must also be demonstrated by the applicant in relation to the existence of harm and a direct causal link between that harm and the infringement in question (para. 54). To be consistent with recital 15 of Directive 2014/104/EU, the required degree of plausibility must not be too strict, as this would weaken the effectiveness of the competition rules (para. 82).

IV. Conclusion

Further to previous cases such as C-25/21 Repsol Comercial de Productos Petrolíferos and C-163/21 PACCAR, the AG’s opinion in case C-286/24 Meliá Hotels International v Associação Ius Omnibus adds more clarity with regard to the disclosure claims under the Directive 2014/104/EU and elaborates the role of trade secrets and the in the context of private enforcement of competition law.

Besides the need to carefully weigh up the interests of all parties involved and the importance of appropriate protective measures for confidential information which, however, have not been discussed in more detail, a decision finding an infringement of competition law is insufficient to establish the plausibility of a claim for damages. Notwithstanding the fact that the decision pertains to a vertical restraint by object and was issued in the context of settlement proceedings, this assessment remains unchanged. At the same time, it once again underscores the role of national courts in interpreting and applying Directive 2014/104/EU in a manner that balances the interests of claimants and defendants by ensuring that the disclosure of evidence in competition law cases is proportionate and adequately protects confidential information. For this, the AG provides clear guidance on proportionality and is setting out the necessary safeguards for the disclosure of evidence, highlighting that the plausibility of a claim for damages must be supported by concrete facts and evidence. This is particularly important as companies are increasingly concerned that the disclosure of trade secrets in the context of cartel damages claims could jeopardise their competitive position.

[1] In its written observations, Meliá argued that Article 5(1) of Directive 2014/104/EU requires a degree of probability that goes beyond the mere possibility of the existence of harm. The AG understood the third question to encompass both the harm and the link between that harm and the alleged conduct and reformulated it to the effect that, by that question, the referring court seeks to ascertain whether, in order to support the plausibility of a claim for damages under Article 5(1) of Directive 2014/104/EU, it must be shown that it is more likely than not that the conditions for liability for an infringement of competition law are met.

[2] The AG interpreted the third question as referring to both the damage and the link between that damage and the alleged conduct. They reformulated the question to clarify that the referring court is seeking to establish whether, to substantiate a claim for damages under Article 5(1) of Directive 2014/104/EU, it is necessary to demonstrate that the conditions for liability for an infringement of competition law are more likely than not to be met.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

I have argued elsewhere that, to do competition assessment in ecosystems, we need at least three additions to the enforcers’ toolkits:

1. “Ecosystem definition”: empirical methods to define the boundaries and structure of an ecosystem. These are based on measuring commonalities across products, as proposed in a joint paper with Konstantinos Stylianou. These methods should supplement traditional market definition, which relies on measuring demand
   substitution between products.
2. Ecosystem mechanisms of harm. Economic theory and evidence on how the ‘usual suspects’ such as economies of scale and scope in data, cross‑platform network effects, or demand steering actually impair competition, and marshal evidence accordingly.
3. Ecosystem theories of harm. New theories of harm that capture non-substituablity-based competitive harms. This post focuses on the third item.

Competition beyond demand substitution in ecosystems

Existing rulings evoking ‘ecosystems’ focus on conventional substitution‑based rivalry, yet ecosystems generate other competitive dynamics. While this is not problematic (of course, firms do compete to sell substitutes within and outside of ecosystems), it is not the whole story. The academic literature shows that, in ecosystems, firms also compete in other ways than by selling substitutes. Firms that cooperate to form an ecosystem also compete for value capture within it, as the Epic Games chapter of the Apple cases saga has vividly illustrated. Today’s complementors might become tomorrow’s competitors, as extensively discussed in the envelopment literature. The collection of assets and capabilities that make up an ecosystem compete with those of a rival ecosystem. Hence, a firm may face competitive pressure or suffer anti-competitive conducts from firms that inhabit different product spaces but the same (or a rival) ecosystem.

Taking these forms of competition beyond (current) substitutability into account should lead to new, ecosystem-specific theories of harm, but also efficiency defenses. Using the Booking/eTraveli case as inspiration, Batra, de Bijl and Klein have proposed theories of harm rationalizing when and how a merger between two firms located in different relevant markets can either increase or decrease ecosystem-vs-ecosystem competition. Yet no agency had operationalized these types of academic insights until CADE’s recent decision on Apple.

CADE’S ecosystem theory of harm

I will not comment on the case in general. Instead, I want to focus on one of the three conducts that CADE investigated, and on which, in my view, the very first ecosystem theory of harm was put forward by a competition authority. Mercado Livre (a leading e‑commerce and payments platform) rolled out an app update offering its ‘level‑six’ loyalty users discounts on HBO GO and Disney+. Apple first objected to the ‘buttons, external links, or other calls to action’ that bypass in‑app purchase. Later, it claimed the feature let users ‘purchase digital content or services for use outside the app.

CADE characterized Apple’s behavior with an envelopment-flavored ‘defensive leverage’ theory of harm. According to this theory, “the restrictions imposed aim to prevent complementor (developers) from disintermediating the central platform or creating alternative distribution channels that could threaten Apple’s monopolistic position”. In other words, Apple’s restrictions aim at impeding Mercado Livre and, importantly, any other complementor (i.e., app developer), from disintermediating the orchestrator (Apple) or “creating alternative distribution channels that could threaten Apple’s monopolistic position”. By this token, Apple would be abusing its orchestrator role in the ecosystem to impede future competition from complementors, regardless of whether they are active in a market in which Apple is active.

Unlike other theories of harm in prior similar cases brought up against Apple, CADE’s theory does not rely on leverage, bundling, or tying. Moreover, it does not focus the harm on a specific market, as other jurisdictions did (music streaming in the EU, dating apps in the Netherlands and gaming in the US), but on the ecosystem structured around the App Store ecosystem taken as a whole. It is essentially a theory of harm about the abuse of market power by the orchestrator on complementors within an ecosystem. In that respect, it is, in my view, the very first ecosystem theory of harm to permeate competition practice.

CADE’s decision could solve the chicken‑and‑egg dilemma ecosystem theories of harm have been facing: enforcers avoid novel theories because there’s no precedent, so no precedent ever arises. To be fair, Apple’s veto of the Mercado Livre update is a textbook example found only in Latin America to date. European and American competition authorities would have had a harder time introducing an ecosystem theory of harm in the cases they had to deal with, regardless of how novelty-(un)savvy they might be. In that respect, CADE’s decision has set the pace for other Latin American jurisdictions. Mercado Libre has filed the same complaint in Mexico, where the Instituto Federal de Telecomunicaciones and the competition authority, COFECE, are reviewing the case. A similar case against Apple is likely to appear in some of the other 16 Latin American jurisdictions where Mercado Libre is a large player in ecommerce and digital payments, notably in Argentina, its founding market and headquarter.

From Promise to Proof: an Interdisciplinary Path to Ground Ecosystem Theories of Harm

While promising and welcome, for the adoption of ecosystem theories of harm to make a difference, authorities must gather evidence on how certain conducts or mergers can strengthen or soften competitive pressures other than (but without forgetting about) demand substitution. The mechanisms through which this might happen should be substantiated. Harder-to-defend counterfactuals will have to be made.

CADE’s decision, grounded in academic work across fields, lights a beacon for collaboration between competition authorities and academic experts. Imagine economists quantifying cross-product complementarities and their impact on ecosystem-wide market power; management scholars assessing organizational and dynamic capabilities to back up counterfactuals; computer scientists showing the effect of recommendation algorithms on product visibility; and psychologists running controlled studies on user choice, each insight feeding antitrust probes. That kind of synergy can transform “ecosystem harm” from a promising idea into empirically grounded theories. Otherwise, ecosystem theories of harm risk being, at best, a passing fad and, at worst, a veneer of enforcement that masks analytical gaps. We must resist the temptation to let a new buzzword justify enforcement without rigorous evidence, just as others have justified inaction.

The views expressed in this post are the author’s personal opinions. They do not reflect the official opinion of the Joint Research Centre or the European Commission. No conflict of interest to declare.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

The Supreme Court of India recently passed its judgment in the Schott Glass case^[1]. The ruling comes after more than ten years when the Competition Commission of India first adjudged the matter and passed a contravention order against Schott Glass India Private Limited (‘Schott Glass’). The SC has mostly confirmed the findings of Competition Appellate Tribunal (COMPAT) setting aside the contravention order and highlighted the commercial justification of the impugned business practices. The judgment has further shed light on policy where it has suggested that India should take advantage of the current geopolitical scenario and practice prudence when it comes to regulating import of foreign technology. The ruling gains importance as Indian competition law jurisprudence is currently in nascent phase and the Supreme Court’s opinion would provide much required guidance to the Competition Commission of India both on law and policy as far as interpreting section 4 of the Competition Act, 2002 is concerned. This is more so when the National Competition Policy is still in the drafting phase. The ruling would further allow commercial enterprises to comply with competition law by balancing pro-competitive and anti-competitive effects of a business practice.

Factual and industrial background

The relevant product in question was ‘Neutral USP-1 Borosilicate glass tubes’. There were five players in the Neutral USP-1 Borosilicate glass tube market out of which three exited due to various reasons. Resultantly, Schott Glass became a dominant entity with Nipro Glass Private Limited being a distant second. The said product is used in two variants of ‘clear’ and ‘amber’ to manufacture respective ampoules, cartridges and vials to be supplied to the pharmaceutical industry. The case was initiated by Kapoor Glass Private Limited (‘Kapoor Glass’/ ‘Informant’) which was a ‘converter’ for the supply of ampoules and cartridges to various pharmaceutical companies.

Kapoor Glass primarily alleged that Schott Glass was abusing its dominance in the upstream market of Neutral USP-1 Borosilicate glass tubes to cement its position in the downstream market of ampoules and cartridges made of such borosilicate glass tubes, a violation of the Competition Act, 2002 (‘Act’). This supply chain between Neutral USP-1 borosilicate glass tube manufacturers, ampoule converters and pharmaceutical companies is important as any competition assessment ought to consider these linkages before reverting a finding. The importance is underlined for pharmaceutical companies who were never consulted during the investigation phase as the final consumers of ampoules and cartridges by the Director General but are now still bound by the court ruling.

Legal analysis

At the outset, the authors submit that there are three orders passed by the Commission (majority order, minority order, supplementary order)^[2] and a judgment passed by the Competition Appellate Tribunal^[3] and Supreme Court each in the case. A critical analysis of these combined orders would suggest that some of the key issues under consideration were the competitiveness of the Neutral USP-1 borosilicate glass tubes and subsequent ampoules and cartridges, contractual terms between the entities in the supply chain and the legal standard for the competition authority to intervene.

The prime text on Indian competition law remains the Competition Act, 2002^[4]. The preamble of the law focuses on ‘promoting competition’ and ‘consumer welfare’ and the three substantive provisions are anti-competitive agreements (section 3), abuse of dominance (section 4) and merger control (section 5-6). Interestingly, while the phrase ‘Appreciable Adverse Effect on Competition (AAEC)’ appears in anti-competitive agreements and merger control provisions, its absence is conspicuous in abuse of dominance^[5]. This aspect is at the core of assessing the ‘cause vs. effect’ debate^[6] of an anti-competitive activity and whether it is even applicable in the context of abuse of dominance, notwithstanding the ambiguity around what are the core elements of an ‘effects-test’.

Another key aspect is the nature of proceedings under Indian competition law. Section 19(1) of the Act makes it very clear that an ‘information’ (vs. complaint) is to be filed before the CCI to set the ball rolling^[7]. This position has been settled by the Supreme Court first in the CCI vs. SAIL case^[8] where the Court held that the Competition Commission of India has broad discretionary powers to grant a ‘right to hearing’ to the party before dismissing a case under section 26(2) of the Act. The Court re-emphasised the regulatory powers of the Commission in Samir Agrawal case^[9] where it held that the proceedings under the Act are in rem in nature. The said regulatory powers gain importance as once a prima facie contravention order is being passed by the Competition Commission of India, there is virtually no bar on the Director General (DG) in terms of defining the methodology to conduct the investigation. As the facts would show, the Director General in this specific case ought to consult the pharmaceutical companies who were the final buyers of the ampoules, the leveraged segment for Schott Glass to conduct a holistic competition assessment.

Out of the total of six orders, two orders stand out. One is the majority order passed by the CCI finding a case for violation and the minority note which was relied upon by the Competition Appellate Tribunal and the Supreme Court while overturning the majority order. The evidence used by the Commission in the majority order was that the overall terms & conditions offered by Schott Glass for the supply of glass tubes to ampoule manufacturers were such to cement its dominance in the upstream market of glass tubes and enter the downstream market of ampoules and cartridges. It further held that the volume-based discounts given by Schott Glass to Schott Kaisha, though per se not anti-competitive, puts the latter in the position of strength through increased profits thereby violating the provisions of the Act.

In this backdrop, some of the observations made in the minority order are of interest. In particular, the dissenting member while scrutinising the discount policy of Schott Glass, at one place, mentioned that it has failed to uniformly apply the discount rates. These comments are important as the gravamen of the dissent lies in commercial justification of the Schott Glass’s (theoretical) discount policy and how the majority order has erred in overlooking the economic aspects of it. The fact that there was divergence between the written text and the implemented text, the judgment ought to hinge on evidence rather than economic theory. The dissenting member goes beyond and notes that the ampoule manufacturers were always at liberty to diversify their procurement options, should they be willing to forego the 8% discount offered by Schott Glass on the condition of exclusivity. It is submitted that such observations are unsustainable both from a legal and economic standpoint as it is a settled position under the law that any starting point for a legal inquiry under competition law is the conduct of a dominant enterprise and not the recipients of it. Similarly, any foregoing of such a discount would be catastrophic to the economic efficiency of ampoule manufacturers which would result in increased input costs.

‘Special responsibility’ under Indian competition law

One of the major jurisprudential values of this ruling is whether there is a ‘special responsibility’ cast upon a dominant entity to comply with the provisions of Indian competition law. The Supreme Court has yet again missed an opportunity to opine on this question which is key to interpreting any factual matrix alleging imposition of unfair and discriminatory conduct by a dominant enterprise. For instance, a monopolistic entity is bound to deal under fair and non-discriminatory terms with its upstream and downstream supply chain partners in the European Union (EU) where such a doctrine is applicable. On the contrary, a dominant entity has higher freedom under the US antitrust law when it comes to deal with its partners^[10]. While the CCI majority order has taken a position on the same, the minority note, Competition Appellate Tribunal ruling and the Supreme Court judgment are silent on this.

In the Fast way transmission case^[11], the Supreme Court looked into the factual matrix and held the respondent (‘Fast way Transmission Private Limited’) in violation of section 4 of the Competition Act. It, however, did not take a position on the special responsibility doctrine. Similarly, in the Coal India Limited case^[12], the Court ruled that an enterprise already governed under the provisions of the Nationalisation Act could be scrutinised by the CCI under section 4 of the Act but kept the question of applicability of ‘special responsibility’ doctrine open. All of this has led to legal uncertainty resulting in the much abhorred ping-pong game between the CCI and – which is now replaced by the National Company Law Appellate Tribunal.

Conclusion

This is the third instance when the Supreme Court has substantially reviewed section 4 of the Indian Competition Act. The Court has practiced an unsought silence on whether a dominant enterprise has special responsibility to comply with the provisions of the Act. This is more so when the CCI has consistently answered this question in affirmative resulting in ambiguity. The second question is whether AAEC is the correct measure when it comes to observing violation of section 4 of the Act, more so when its absence is conspicuous in the legislative text. It is high time that the Indian government takes appropriate legislative measures to address this gap so that concepts such as ‘legal certainty’ and ‘ease of doing business’ are felt for real, rather than remain merely an academic nicety.

The authors are thankful to Dr. Christian Bergqvist, Professor of Law, University of Copenhagen for his comments on the earlier draft.

^[1] https://api.sci.gov.in/supremecourt/2014/19707/19707_2014_5_1501_61745_Judgement_13-May-2025.pdf

^[2] https://www.cci.gov.in/antitrust/orders/details/877/0

^[3] Competition Appeal (AT) No. 91 and 92 of 2012

^[4] The Competition Act, 2002, No. 12, Acts of Parliament, 2003 (India)

^[5] Supra

^[6] Payal Malik et al., Legal treatment of abuse of dominance in Indian competition law: adopting an effects-based approach, 54(2) Review of Industrial Organisation 435, 436 (2019).

^[7] Supra

^[8] Competition Commission of India vs. Steel Authority of India Ltd. (2010) 10 S.C.C. 744

^[9] Samir Agarwal vs. Competition Commission of India (2021) 3 S.C.C. 136

^[10] Centre for Competition Law and Economics, Memorandum of the Panel Discussion on “Competition in Digital Markets: A Look at the US AdTech Trial” (Nov. 25, 2024),

https://www.icle.in/wp-content/uploads/2024/11/US-AdTech-trial-conference-memorandum.pdf.

^[11] Competition Commission of India vs. Fast way Transmission Private Limited, (2018) 4 S.C.C. 316 (India)

^[12]  Coal India Limited vs. Competition Commission of India, (2023) 10 S.C.C. 345 (India).

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

While the TCA’s decision concerns digital markets in Turkey, its importance may go far beyond that. Competition authorities and policymakers are seeking the most effective approach to address potential innovation concerns arising from mergers. For example, the EU initiated a review process for its merger guidelines, upon the recommendation in the Draghi report. The review process includes in-depth consultation on seven specific topics. Two of these topics, Topic C on innovation and Topic E on digitalisation—explicitly mention issues such as killer acquisitions and innovation considerations in merger control.

Big Tech’s Hunger for Mergers

It is a challenge to find the exact number of acquisitions Big Tech has made so far. Parker, Petropoulos and Van Alstyne find that Big Tech (Google, Amazon, Apple, Microsoft and Meta) made 855 acquisitions between 1998 and 2020, with approximately 97% of these mergers having gone under the radar, according to Kwoka and Valetti. More recently, another recent research and a public dataset by SOMO shows that Big Tech acquired 191 companies between 2019 and 2025, meaning that Big Tech acquired a company every 11 days in the last 6 years. One thing is for sure: mergers are one of the main business strategies of Big Tech firms.

Authorities have been unable to intervene so far, primarily due to most of these mergers falling below notification thresholds and a limited understanding of how competition works in digital markets. Moreover, the prominence of non-price factors—such as innovation, quality, interoperability, and privacy—has made it difficult to assess the true impact of these mergers.

A different approach to below-threshold mergers: Technology Undertakings

Like most jurisdictions, Turkey’s merger control regime was (and still mostly is) based on the annual turnovers of the merging parties. That said, in 2022, with the Communique No. 2022/2[2], the TCA brought a novel concept of technology undertakings, creating an exception to the turnover-based merger control regime. Technology undertakings are defined as undertakings and their assets operating in specific industries: digital platforms, software and gaming software, fintech, biotechnology, pharmacology, agriculture chemicals and health-tech.

Accordingly, if the target technology company:

• Operates in Turkey or
• Conducts R&D activities in Turkey or
• Provides services to users in Turkey

Then, merging parties must notify the transaction to the TCA, as long as the acquirer’s annual turnover exceeds the notification thresholds. This is quite a different approach from a deal value-based threshold (such as in Germany) or the call-in power approach in multiple European countries (such as Italy).

(Substantial) Activity in the Domestic Market

The assessment of the TCA is twofold (i) whether the proposed merger is subject to mandatory notification, and (ii) whether the merger significantly impedes effective competition. The identification of Galileo as a technology undertaking is pretty simple. According to the TCA, Galileo qualifies because it provides software-based technology services. As Google’s turnover exceeds the relevant notification thresholds, the transaction is subject to mandatory notification (para 11). Contrary to other jurisdictions, such as Germany, the technology undertaking approach doesn’t require the authority to find substantial activity of the target company in the domestic market to review the merger. Although the amendment mentions that the target company must operate or have R&D activities or provide services to customers in Turkey to be reviewable by the TCA, there is no requirement for the extent of the activity. Therefore, even though Galileo had limited activity in Turkey (para 25), the existence of customers in Turkey (para 17) was sufficient for the TCA to review the merger.

The substantial activity requirement limits competition authorities from intervening in cases dealing with the acquisition of a small company. Indeed, because of this requirement, the Higher Regional Court of Dusseldorf stated that the Bundeskartellamt doesn’t have jurisdiction to review Adobe’s acquisition of Magento and Marketo as the target companies didn’t have substantial activity in Germany.

Similarly, call-in power was subject to criticism by companies. In October 2024, the European Commission accepted the referral request from the Italian Competition Authority (AGCM), which used its call-in power to review the Nvidia/Run:ai merger. The merger was later approved by the European Commission. However, Nvidia filed a lawsuit (Case T-15/25) against the European Commission’s acceptance of the referral from AGCM, alleging that AGCM’s exercise of “loosely defined, ex-post, discretionary call-in powers infringes institutional balance, legal certainty, proportionality, and equal treatment.

In these two aspects (domestic activity and discretionary powers), the technology undertaking approach has certain advantages in capturing below-threshold mergers, although there might be other concerns related to this approach.

Assessment of Competition

After holding that the merger falls within the scope of the Turkish merger control regime, the TCA analyses the competition in relevant markets. The genAI tools that Galileo develops enable users to generate high-quality designs by entering text and image inputs.  Google stated in the notification form that it intends to integrate Galileo AI into its Google Labs, a platform that allows developers to test experimental AI products. Further, according to the notification form, although Google Labs has an unreleased tool for UI design, Galileo’s contribution to developer tools would be minimal, given the deal value and Galileo’s limited turnover (para 22).

The decision first identifies Google’s activity in the AI development and mentions its presence in cloud (Google Cloud), foundation models (such as Gemini), AI development platform (Vertex AI), and genAI applications (Gemini chatbot, Gemini-integrated Google Workspace). The decision states that Galileo didn’t develop its own foundation model (FM) but used third-party FMs to develop its UI design tools (para 24). Since Google doesn’t offer UI design tools, TCA concludes that there is no horizontal overlap.

That said, TCA finds a vertical overlap between the activities of Google and Galileo (para 25). Unfortunately, the specific details of the vertical overlap and the potential concerns it raises have been redacted from the public document to protect commercially sensitive information. The sentences that follow the redaction seemingly suggest that the primary concern is related to the vertical overlap between the Google Gemini model and Galileo’s generative AI tools—specifically, the risk of potential foreclosure. However, TCA concludes that such a foreclosure is unlikely to produce anticompetitive effects due to (i) Galileo’s limited activity in Turkey and (ii) existence of alternatives of Google’s Gemini model such as Microsoft/OpenAI (interestingly the decision mentions these two companies together), Meta and X (para 25). Therefore, the TCA finds that while the extent of a startup’s activity in Turkey is irrelevant for determining the applicability of Turkish merger control, it is taken into account when assessing the substance of the case.

What are the conditions of a killer acquisition?

The decision then includes a risk assessment related to the killer acquisition theory. In that end, the TCA identifies three criteria for killer acquisitions (para 34):

1. Acquisition of a startup by a scaled-up incumbent
2. The acquired product or technology not being adopted, developed, maintained, or being entirely withdrawn from the market and as a result,
3. Elimination of horizontal competition and product development process

The second criterion is particularly interesting, providing a more helpful interpretation of the killer acquisition theory, as it can be applied not only when the acquirer shuts down the target, but also when it neglects it. For example, this interpretation can be well-applied in the FTC v. Facebook case, in which the Instagram co-founder said Meta starved Instagram of resources after being acquired.

After applying this interpretation to the case, the TCA concludes that there is no risk of a killer acquisition because (para 35)

• Google will integrate Galileo into Google Labs
• There is no horizontal overlap between the activities of Google and Galileo and
• Galileo competes against other products such as Adobe XD and Figma Design, which have the same technology.

As a result, TCA decides that while the Google/Galileo merger is subject to merger review, it doesn’t significantly impede effective competition.

Diverging Opinions

However, the story doesn’t end there. 3 out of 7 members of the Turkish Competition Board dissented from the decision to approve the merger. The dissenting opinion criticizes the TCA’s analysis for its shortcomings, arguing that the Google/GalileoAI transaction, given Google’s dominant position across the AI value chain, has the potential to be a killer acquisition. It emphasizes that the merger could significantly hinder competition through its vertical integration and ecosystem effects.

The dissent criticizes the analysis as it merely mentions academic literature and does not thoroughly analyze the potential harms to competition. The dissent highlights that digital ecosystems—such as the Google ecosystem centered around Google Search—are largely shaped through mergers and acquisitions. It also draws on the UK Competition and Markets Authority’s (CMA) report on foundational models (FNs), noting that entry into AI markets is limited due to bottlenecks.

In this context, the dissent argues that Galileo’s text-to-UI and image-to-UI tools are significantly important, but the TCA’s analysis fails to conduct a substantive competitive assessment. Instead, it merely lists the names of the competitors without further analysis. Further, the dissent criticizes the analysis as it didn’t scrutinize whether Galileo’s products and services are strategically important in certain sectors.

The dissenting opinion draws on precedent cases—Illumina/Grail, Visa/Plaid, EEX/Nasdaq Powerdeals, and Qualcomm/Autotalks—to argue that the merger warranted a more thorough assessment, particularly regarding potential foreclosure risks, adverse effects on innovation incentives, labour market implications, the AI ecosystem, and the identification of potential competition restrictions across all relevant markets.

Finally, the opinion refers to another Google merger reviewed by the FTC—Google’s acquisition of AdMob—where Google opted to buy a technology it could have developed in-house. The dissenting opinion states that had the merger not been approved, Google might have been incentivized to build its own tools, while AdMob would have had stronger incentives to further invest in its own technology.

In summary, the three members of the TCA Board objected to the decision, stating that competition assessment in innovation-centered markets differs from that in traditional markets, and that the TCA’s analysis did not take a holistic approach in this regard.

Conclusion

This decision is particularly insightful, as it demonstrates that the assessment of novel concepts, such as killer acquisitions, should include a detailed analysis of the incentives and capabilities of merging firms to drive innovation.  Such an analysis should not only be limited incentives or intentions of the merging firms, but it should take lessons from previous cases to evaluate the aftermath of acquired technologies.

The decision provides two important lessons for anyone interested in merger control in digital markets. Turkey is an interesting example when it comes to tackling mergers where innovation is particularly important, and it offers an alternative to other rules, such as call-in power and deal value-based thresholds. Considering the EU is struggling to find a uniform solution to killer acquisitions, there might be lessons to take from the Turkish experience. Secondly, it also shows the diverging opinions within competition authorities when it comes to tackling digital mergers. The willingness to challenge Big Tech’s dominance remains strong in jurisdictions beyond the US and EU, highlighting the global anti-sentiment toward their power.

[1] Turkish Competition Authority decision number: 25-02/62-37, published on 03.06.2025

[2] Rekabet Kurulundan İzin Alınması Gereken Birleşme ve Devralmalar Hakkında Tebliğ (Tebliğ No: 2010/4)’de Değişiklik Yapılması Hakkında Tebliğ (Tebliğ No: 2022/2)

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

Key considerations and concerns revolve around high barriers to market entry, network effects and tipping, integration of foundation models into the larger ecosystem of digital gatekeepers and dependency on large digital market players for essential infrastructure to train generative AI models. The UK Competition and Markets Authority (CMA) studied competition concerns along the foundation models (FM) value chain, and acknowledged competition concerns such as self-preferencing, vertical integration, refusal to supply and unfair terms and conditions. The French Competition Authority (FCA), Autorité  de la concurrence, assessed competition concerns at the following three layers in the generative AI value chain: the upstream layer, as in the AI infrastructure layer; the middle layer, i.e. the AI modelling layer and the AI deployment layer. This discussion follows this three-layered approach to assess competition concerns in the generative AI value chain and offers an overview of the key insights and research findings of the pre-print “Mapping competition concerns along the generative AI value chain” available here.

The upstream infrastructure layer

In the upstream market, the key inputs are computing power, data and skilled workforce. In addition, availability of venture capital and financial resources is also essential to a vibrant generative AI market.

Considering the large sunk costs involved in developing generative AI models, a vibrant and well-functioning capital market is critical to a local innovative ecosystem. In fact, a key reason for the thriving generative AI market in Silicon Valley is the abundant availability of venture capital (VC) to fund promising start-ups.

Flow of VC funding is also an indicator of the potential and promise of an emerging market. As Google became the dominant search engine, investment in search engines virtually came to a halt. The market for funding R&D and commercialization in generative AI, on the other hand, saw a geometric increase. Between 2022 and 2024, funding in generative AI-based start-ups quadrupled to reach $25.2 billion in 2024. While it is true that financial funding is the lubricant that helps run smoothly high sunk cost-driven innovative projects, considering that this limitation in the upstream layer cannot be remedied through competition and regulation, the follow-on discussion revolves around the three other key inputs that are sine qua non for generative AI. These include: computing power, data and talent acquisition (or acqui-hires).

Computing power

Central Processing Units (CPUs) are the brain of our digital devices, such as laptops and computers. A key limitation with CPUs is that they cannot cope well with the complexity of large language, deep learning models. To resolve this, graphic cards, used mainly for gaming, emerged as a promising alternative. Graphic processing units (GPU) could work well with complex, high-powered processing required in training large language models. Thus, the rise of big data revolution followed by improvements in hardware capabilities, including innovations in GPUs, led scientists to pro-actively engage in neural network-based approach to training generative AI models.

Nvidia was an important contributor to this new wave of innovation. As ChatGPT, one of the fastest and most widely adopted app in digital history, trained on Nvidia’s offerings, the latter’s valuation quickly rose from $300 billion in November 2022 to $2.3 trillion in 2024. From an innovation perspective, this is an important consideration, as Nvidia is a remarkable contributor to the uptake of the generative AI revolution. However, it has also been under the antitrust radar for its exclusionary practices in the sale of GPU chips, as well as its acquisition of Run:ai, an AI optimization software.

Nvidia’s capabilities are complemented by other key players along the manufacturing value chain. For example, it outsources manufacturing to Taiwan Semiconductor Manufacturing Company (TSMC) and Samsung Electronics. TSMC is, in turn, dependent on the Dutch national champion, ASML, that is the world’s ‘only’ supplier of extreme ultraviolet (EUV) lithography machines, that is an essential input to manufacture GPU chips. To develop an ecosystem around its core offering of GPU chips, Nvidia has developed a software, Computer Unified Device Architecture (CUDA) that helps developers simultaneously use processing power from different sources.

To further strengthen this software capability, Nvidia also recently acquired Run:ai, a so-called killer acquisition. Nvidia/Run:ai merger was first caught by the Italian Competition Authority (ICA), L’Autorità Garante Della Concerrenza e Dell Mercato (AGCM) under its special call-in powers. As per Article 16(1-bis) of the Italian competition law, three conditions must be cumulatively met for a proposed transaction to qualify for review: first, the notification to the merged entity must follow within the first six months of the consummation of the transaction; secondly, either the merger should meet the turnover-based thresholds under Article 16(1) or alternatively, the merged entity must have an annual worldwide turnover of €5 billion or more and thirdly, there must be an identifiable risk to innovation in part or entirety of the relevant market. As the merger met all the three conditions, the ICA referred the merger to the European Commission under Article 22(1) of the EU Merger Control Regulation. Following a phase I investigation, the Commission, unconditionally approved the transaction.

Unlike Nvidia, however, the other players in the market, namely TSMC and ASML, remain focussed on their core capabilities, and have not developed an ecosystem around their main offering. Nvidia, with over 90 per cent market share, is a monopolist in the market for GPU chips. However, it is not without competition. Key sources of potential competition include Google, Amazon and Microsoft, that are investing heavily in developing their own GPU chips. In addition, Meta is working on its eAccelerator chips and IBM is working on its flagship, Telum project to offer a viable competitive alternative to Nvidia’s GPUs. Considering that multiple competition projects are being pursued to develop a better GPU, any one of these projects, if successful, may in the long run, prove to be a workable competitive alternative to Nvidia’s GPUs.

While Nvidia’s GPU may face some emerging competition, the market for cloud computing is dominated by the three cloud hyperscalers, namely Amazon Web Services, Microsoft Azure and Google Cloud Platform. In light of the findings from its cloud market study on the position of strength enjoyed by these hyperscalers, the UK telecommunications regulator, OfCom, requested a full market investigation by the UK Competition and Markets Authority. As noted by the French Competition Authority, and the UK Competition and Markets Authority (CMA), the market for cloud computing is highly concentrated on account of substantial costs involved in building a data center, technical barriers to customer migration and data integration, egress fees, committed customer spend discounts and restrictive licensing practices. The highly concentrated nature of digital markets and cross-linkages between different services in the ecosystem helps hyperscalers further fortify their position of strength. Increased demand for GPUs and cloud computing following the rise and the mainstream adoption of generative AI is only expected to strengthen the position of hyperscalers. This can be attributed to the fact that developing, running and maintaining generative AI models requires large computing resources.

Open AI, one of the notable unicorns in generative AI, benefited substantially from its close collaboration with Microsoft. In addition to the $12 billion funding that OpenAI received from prominent Silicon Valley based investors, such as Microsoft, Khosla Venture, and A16Z, Microsoft also developed a dedicated cloud infrastructure for OpenAI to train its GPT models. Following this close collaboration in the infrastructure and modelling layer, OpenAI’s products are exclusively available on Microsoft Azure. Thus, this collaboration has spill-over effects across the entire generative AI value chain. Considering the centrality of cloud infrastructure and the position of strength enjoyed by these hyperscalers, an important consideration is whether the Digital Markets Act (DMA) can offer relief. The consideration here is not whether generative AI can be captured by the DMA (an issue that I turn to later in the discussion), instead the question here is whether competition and contestability in cloud computing, that is ‘an essential infrastructure’ for generative AI, can be facilitated through the DMA. Trite to add here that Article 2(2)(i) of the DMA refers to cloud computing as a core platform service (CPS). The hyperscalers, Google, Amazon and Microsoft, have already been designated as gatekeepers under the DMA. Currently, however, their cloud computing services are not qualified as CPSs as they do not meet the prescribed thresholds under the DMA nor has the EC decided to trigger designation based on qualitative criteria.

Bearing in mind the centrality of cloud to the digital and now the generative AI infrastructure, it may be important to re-consider whether cloud services can be caught by and subject to the obligations under the DMA. The market investigation tools as available under Chapter IV of the DMA may be a useful starting point to evaluate this possibility. Instructive in this regard is also a competition law-related development, wherein Google complained to the European Commission that Microsoft was engaging in anti-competitive tying by forcing customers to select its cloud service, Azure alongside its main offering, Windows Server.

Data

Data is the food that feeds generative AI models. Data is thus, a key input in the training of generative AI models. Google, Amazon, Microsoft, Meta and Apple (GAMMA) control rich sources of valuable data. Google has access to the data generated by YouTube, Google Search, Gmail, Google Maps and its Android App Store. Meta has rich sources of personal communication and content generated data accessible on its social networking and communications apps, namely, Facebook, Instagram and WhatsApp. Amazon has access to users purchasing habits on Amazon.com, as well as data generated over and for its services, such as Prime, Twitch and Alexa. Microsoft has access to data generated on its productivity software, and communication apps, such as Teams. Apple has access to data on Apple Watch, Siri and its App Store.

In addition, considering their position of strength, large digital firms may also tend to benefit from the data feedback loop. The use of these models leads to more data input, which leads to a better product, and as this data is used to train the model, a better product leads to a larger user base, which in turn again offers more data, and the cycle continues perpetually.

Generative AI has . It was trained using ‘300 billion words’ scrapped from the worldwide web. This content included both personal as well as non-personal data, without the consent of the rightsholders or the data subjects. This draws a close parallel with Google’s advantage as the world’s most popular search engine, which, in turn, offered it an unparalleled advantage while training its generative AI model Gemini (formerly, Bard). Google derives unparallel advantages from being the world’s most popular search engine. It receives over 93 per cent of unique queries, when compared to  3.8% queries received by its closest competitor, Bing. In other words, thanks to its unique access to queries and data, Google’s algorithms can make more meaningful inferences about the relevance of a given query. Likewise, generative AI models, such as ChatGPT, have trained principally on web-scrapped data. Moreover, to restrict follow on access to data to other small and medium enterprises, ChatGPT has restricted access to this proprietary data generated on its platform.

Data can be of many different types. When original and a result of author’s own intellectual creation, data source may be subject to copyright and related rights. From that lens, the outcome in the ongoing copyright related lawsuits in the US, EU and the UK, have important implications for competition and innovation in the market for generative AI models. In the EU, the licensing framework available under the 2019 Copyright in the Digital Single Market Directive (CDSM) is seen as an enabler to ensure lawful access to copyright-protected proprietary data for training generative AI models (a process known as text and data mining).

Thus, lawful access to copyright-protected data is not only about fair and proportionate remuneration of the romanticised human author that sits at the core of copyright; the issue is equally central to facilitating contestability in generative AI.

In case gatekeepers have a data advantage that flows from their CPS (such as Google’s data advantage from Google Search), access can be enabled via the DMA. Article 6(10) of the DMA requires authorized ‘effective high-quality, continuous and real-time access to, and use of, aggregated and non-aggregated data, including personal data’. Such access must be provided in compliance with the consent requirements as set in the 2016 General Data Protection Regulation (GDPR). In addition, Article 6(11) requires access to ‘ranking, query, click and view data’ by the gatekeeper on ‘fair, reasonable and non-discriminatory’ terms. A stringent application of these provisions (on this also see, the Google compliance reports) can be an enabler of competition in training generative AI models. This may be an important enabler, as a dichotomy exists relating to training and control over data. While digital startups are data hungry, the rise of large language models and synthetic data has diminished Google’s (and other digital gatekeepers) data dependency. Google’s older algorithms required over a trillion examples to make meaningful inferences, whereas its more recent algorithms (powered by generative AI-driven advances such as word vectors and transformers) require only one billion examples to make a meaningful inference. Thus, while entrants in the generative AI sector remain data hungry due to lack of economies of scale and scope, incumbent digital players, such as Google and Microsoft (also consider their close partnership with OpenAI), not only sit as gatekeepers on large reservoirs of data, but there enhanced algorithms are also increasingly data efficient. Additionally, the design of smarter algorithms is facilitated by talented technical teams.

Talent acquisition (or acqui-hire)

Talented technical personnel are key to innovation in generative AI. In 2024, two Nobel Awards were awarded in physics and chemistry to researchers working in machine learning, artificial neural networks and the application of machine learning in drug discovery.

In 2021, over 65% of Machine Learning/AI researchers were hired by the industry. Such hiring though legally permissible, it can under circumstances fall foul of merger control rules, particularly when the acquirer acqui-hires the key personnel, such that it offers him de facto control over the acquired undertaking. The Microsoft/Inflection merger presented one such situation wherein Microsoft hired the key personnel from Inflection such that effective ownership was passed into the hands of Microsoft.

The UK CMA assessed the merger as it was deemed a relevant merger situation under the Enterprise Act, 2022. Following a phase-1 investigation, the CMA did not refer the transaction for further assessment, as the merger did not raise competition concerns in the market for the development and supply of foundation models. The European Commission initially accepted the referral from the EU Member States under Article 22 of the EU Merger Regulation. However, following the CJEU’s decision in Illumina/Grail that Article 22 was ‘not a corrective mechanism to capture concentrations’ that escaped national merger control thresholds, the Member States withdrew their referral request and the Commission chose not to proceed further with its assessment of the Microsoft/Inflection deal. Mergers such as Microsoft/Inflection and Nvidia/Run:ai highlight a ‘threshold’ gap in the current EU merger control framework as this is the first filter that determines the applicability of merger regulation.

The AI modelling layer & the AI deployment layer

Following the availability of key inputs, the next level is training in foundation models. Training an AI model is a complex and energy-intensive process. Quality of training and output therein depends on the quality of available inputs at the infrastructure layer (as discussed above).

The trained models can be either general-purpose, or they can be vertically specialized foundation models targeted at certain niche segments in the industry. A GPAI, if available in open source, can be fine-tuned to perform particular tasks, using a process known as ‘domain or task specific fine tuning’. ChemBERTa, for example, is built on Meta’s LLama-2 models and is a vertical FM model targeted at chemistry, cheminformatics, and drug discovery. Considering the targeted nature of these vertical models, they perform better in their targeted niche compared to the GPAI.

Building a general-purpose AI model from scratch is an expensive process. As per estimates, developing a GPAI from scratch can cost up to US$8.8 trillion. GAMMA’s FM models are more general purpose and can be trained for deployment across a range of verticals. These models can be open-source or close source. While a large majority of models, at least at the moment, are available open source, it may be important to consider that these firms do not follow the open early, closed later strategy.

Following the foundation layers of AI infrastructure and AI training, comes the deployment layer. At the deployment layer, AI services are offered to the end user. Microsoft’s Co-Pilot, Meta’s AI-based answer engines on Facebook, Instagram and WhatsApp are some examples of the integration of foundation models into existing digital ecosystems to enhance user output.

Timely application of the DMA can be a key enabler of competition at this level of the value chain. There are two possible approaches to capture generative AI under the DMA. A long-drawn approach may be the possibility to amend the DMA and offer an additional category of CPS looking only at the generative AI models. An alternative, and perhaps more efficient approach, as also illustratively emphasized by Coeuré, Ribera Martínez, Picht and Hoppner, may be to capture generative AI as an embedded feature in the gatekeeper’s already-designated core platform services. A long drawn path to amend the DMA may unsettle the enforcement process, particularly in light of the increasing geopolitical considerations in competition and regulatory enforcement. On the other hand, an effective and efficient use of the provisions of the existing regulatory framework, may not only align with recommendations of the Draghi Report to streamline enforcement, but it may in addition, also facilitate a timely, proportionate and effective intervention in generative AI.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

The EU Foreign Subsidies Regulation (FSR) was initially designed as a legal tool to ensure that foreign state support does not distort competition in the EU internal market. It seeks to scrutinize subsidies by non-EU countries similar to state aid provided by EU countries. That was the idea. However, recent developments suggest attempts to use – or misuse – the FSR as a political shortcut rather than a neutral procedural legal tool.

The Czech Nuclear Case

The recent Czech nuclear tender provides a case in point, as Korea Hydro & Nuclear Power (KHNP) must have been wondering just that over a nuclear power plant construction tender issued by the Czech Republic (worth EUR 16b) it had thought won. Initial reports mentioned that the European Commission, via DG GROW, did not raise any concern on the bid under the FSR. The Koreans, it appeared, were playing by the rules, seemingly on track for their first-ever foray into Europe.

But not everyone was satisfied.  Électricité de France (EDF), thwarted local player in the Czech nuclear procurement and unabashed European champion, reportedly escalated its complaints directly to the European Commission’s President Ursula von der Leyen. This overrode the entire formal FSR review process including Commissioner Stephane Séjourné – a Commission Executive Vice-President, no less – who is in charge for the FSR portfolio. If successful, the message of this action would have to be that if the legal FSR review doesn’t go your way, you go political.

This raised crucial questions about due process. Is the FSR being applied within the checks and balances originally foreseen? Or is it at risk of being reinterpreted on the fly, depending on who is doing the asking?

A Regulation under strain

The Czech nuclear tender is not the only case where the FSR is making a difference. The FSR has triggered far more reviews than expected; DG GROW reportedly processing more than 1,000 public procurement files. And, also on the transactional side, the FSR’s reach is already producing significant behavioral changes. Companies now withdraw or avoid participation in tenders to steer clear of potential FSR complications.

Therefore, a tool meant to ensure fairness could increasingly be seen as a deterrent, particularly for non-EU bidders, if the process can be overridden by direct political appeals.  That, in turn, would undermine both legal certainty and trust in the Commission’s procedural integrity.

Strategic autonomy or strategic ambiguity?

As the Czech nuclear tender also highlights, the FSR process can quickly stray into the wider geopolitical landscape.  The EU wants to foster strategic autonomy and green investment (e.g. in nuclear energy), and at the same time strengthen its strategic autonomy. But what are the criteria for choosing between a Korean supplier and a European champion like EDF?

That can be a fine line to tread.  If FSR enforcement becomes a cover for political preference, legal certainty and market trust suffer. If the Korean bid adheres the market principles and passes the legal threshold, then the case for disqualifying it becomes far less compelling. Using FSR to shield EU powers would not only extend its mandate but risk undermining its legitimacy.

Fortunately, in the Czech nuclear tender, the Commission and the Czech government chose to uphold the process. Mindful to avoid a breach a duty of “sincere cooperation” during an investigation into potentially distortive foreign subsidies, the Czechs signed the deal with KHNP after their Supreme Administrative Court lifted the interim injunction. President von der Leyen was not seen to pick up the undue European champion mantle pitched by EDF, and instead the European Commission meanwhile stated that its preliminary investigation under the FSR remains on-going, but with a strong message that procedural integrity remains a priority.

A defining moment for the FSR

This may well be a defining moment for the FSR. If the EU wants to remain an open economy while managing the geopolitical shifts, it must ensure that the FSR operates as intended: as a neutral, rules-based system. Political interventions that bypass formal procedure undermine this system and risk turning the FSR into a credibility trap.

The real test of the FSR isn’t whether it blocks or clears any particular deal. It’s whether it does so fairly and transparently. This includes a clear boundary between law and politics: when legal instruments are used to serve political ends, the distinction between strategic autonomy and strategic protectionism becomes dangerously unclear.  The Czech nuclear tender experience is encouraging in this regard.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

Before examining Advocate General Kokott’s Opinion in detail, it is necessary to first outline DG COMP’s Decision in Google Android (Case AT.40099 – Google Android), with particular attention to the rationale DG COMP identified as underpinning Google’s abusive conduct. In this case, DG COMP found that, since January 2011, Google had engaged in abusive practices concerning the Android Operating System for smartphones. These practices comprised three elements:

1. Access to the Play Store – used for downloading applications to Android smartphones – was contractually tied, through the Mobile Application Distribution Agreement (MADA), to the pre-installation of (i) the Google Search app and (ii) the Google Chrome browser. This arrangement conferred a competitive advantage on both applications at the expense of rival offerings. This constituted the first abuse of tying, comprising two distinct infringements, and was upheld by the General Court.
2. The licensing of both the Play Store and Google Search was made conditional upon the conclusion of an Anti-Fragmentation Agreement (AFA), which restricted (i) modifications to, (ii) fragmentation of, and (iii) the distribution of software derived from the Android operating system. This constituted the second abuse of tying, which was likewise upheld by the General Court.
3. Under a Revenue Sharing Agreement (RSA), Google made payments to smartphone manufacturers and mobile network operators to incentivize them not to pre-install competing search applications. This practice constituted exclusive abuse; however, the General Court did not uphold it as a separate infringement, but did accept it as contributing to the overall abuse.

Although the restrictions in question were formally distinct, they were closely interconnected. First, all originated from Google’s strategic use of the Android Operating System (OS) and were implemented with the overarching aim of safeguarding its dominance in general search services and protecting the substantial revenue generated from search advertising (Case AT.40099 – Google Android, recital 1341). Second, participation in the Revenue Sharing Agreement (RSA) was contingent upon entering into the Mobile Application Distribution Agreement (MADA), which governed the pre-installation of applications. In turn, the MADA was itself conditional upon acceptance of the Anti-Fragmentation Agreement (AFA) (Case AT.40099 – Google Android, recitals 155–191).

These interlinked practices were treated as a Single and Continuous Infringement – a legal concept that allows multiple, seemingly discrete abuses to be treated as one ongoing infringement. Furthermore, the General Court found that Google had pursued these policies irrespective of a full appreciation of their potentially anti-competitive effects (Case T-604/18 – Google (Android), paras 841 and 1043–1051). On this basis, the Court upheld the imposition of a record fine, reducing the amount only to €4.12 billion to reflect its finding that the RSA, considered in isolation, did not constitute a distinct abuse.

To properly assess the legal and economic principles established by this case, it is first necessary to examine Google’s broader business model and the specific practices that were condemned in DG COMP’s Decision.

1. Google Business Case and Google Android

Google’s core business centers on the operation of a two-sided search platform, which enables users to access internet search services free of charge, while offering advertisers the opportunity to display their advertisements alongside search results or on web pages linked to those results. On the user-facing side, Google provides general search results in response to queries, delivered via its search engine and associated web browser. Although users do not pay for these services, Google derives significant value from the interaction by collecting user data, which both enhances its search algorithms and is monetized through targeted advertising on the advertiser-facing side of the platform.

As part of this two-sided platform, Google has developed a broad ecosystem of complementary services – including Google Maps, Google Chrome, Gmail, YouTube, and hundreds of other applications – often provided at no cost. This open ecosystem strategy, designed to maximize user reach and engagement, distinguishes Google from companies such as Apple, which follow a vertically integrated business model wherein key technologies and services (e.g., Operating Systems) are typically not licensed to third parties.

Separate from its search platform and associated ecosystem, Google also offers Android, an Operating System (OS) for smartphones and other mobile devices. Android enables end users to operate both hardware and installed applications. Smartphone manufacturers and mobile network operators may choose to use Android as a lower-cost alternative to proprietary operating systems. Although Google did not invent the smartphone, the smartphone OS, or the Android OS itself, it acquired the Android platform in 2005 and began distributing it as free, open-source software from 2007. This decision generated significant interest and substantially increased the OS’s early adoption and global rollout.

Despite Android’s open-source nature, Google retained substantial influence over its development through a system of authorizations, technical requirements, and compliance monitoring. These mechanisms enabled Google to ensure a consistent level of quality and interoperability across the Android ecosystem, thereby reinforcing its strategic interests and commercial objectives.

1.1 Google conditioning for installing apps

As part of the Android ecosystem, Google offered a suite of applications known as Google Mobile Services (GMS), which could be installed free of charge by device manufacturers and other partners, provided they entered into a Mobile Application Distribution Agreement (MADA). Under the terms of the MADA, the entire GMS package, including Google Search, Google Chrome, and the Play Store, was required to be accepted in full and prominently displayed on the device’s home screen.

Access to the MADA, however, was conditional upon the acceptance of an Anti-Fragmentation Agreement (AFA), which prohibited the use of modified versions of the Android Operating System – so-called Android “forks.” In addition, Google offered financial incentives under a Revenue Sharing Agreement (RSA) to encourage manufacturers and network operators to pre-install the full suite of Google applications provided through GMS. In principle, this pre-installation was optional, as the Google applications were not technically required for the functioning of Android smartphones (Case AT.40099 – Google Android, recitals 282 and 1092–1113).

Nonetheless, from Google’s perspective, pre-installation was strategically crucial. With the growing dominance of mobile devices as the primary access point to the internet – surpassing desktop use since 2015 – ensuring that its applications were pre-installed on smartphones became central to maintaining its position in the search and advertising markets (Case T-604/18 – Google (Android), para. 149). DG COMP’s case against Google thus implicitly rests on the recognition that this shift in user behavior motivated Google’s investment in the Android Operating System and its subsequent efforts to secure widespread adoption through aggressive distribution practices.

2. DG COMP had relied on a narrow market approach

In its Decision (Case AT.40099 – Google Android, recitals 73–104), DG COMP identified a distinct market for the a) licensing of smart mobile Operating Systems. As further elaborated in recitals 218–267 of the same case, this market was considered separate from: (i) operating systems for stationary personal computers (PCs); (ii) basic (non-smart) mobile Operating Systems; and (iii) non-licensable Operating Systems. Accordingly, the relevant market was defined as encompassing all licensable Operating Systems for smartphones and tablets.

Significantly, and much to Google’s dissatisfaction, Apple’s iOS was excluded from this market definition, as Apple does not license its Operating System to third parties, reserving its use exclusively for its own devices. In practice, Google was the only major provider of licensable mobile operating systems, rendering its dominant position in the market largely inevitable.

Other involved markets was the market for b) Android app stores (Case AT.40099 – Google Android, recitals 268–322), that enable users to download, install, and manage mobile applications on Android devices, and c)  general search services, (Case AT.40099 – Google Android, recitals 323–366), which enable users to search for information across the entire internet, regardless of the device (PC or mobile) or access point used.

In terms of geographical scope, all markets, except general search services (considered national due to language reasons), were accepted as international, excluding only China, as Google has limited operations there (Case AT.40099 – Google Android, recitals 400-430).

2.1 Google was unhappy with the narrow market definition.

Unsurprisingly, Google was unhappy with DG COMP’s narrow definition of the market for licensable Operating Systems, which excluded Apple’s iOS. Google argued that consumers make choices at the downstream level of smartphone retail, and that these choices indirectly shape competition in the upstream market for mobile Operating Systems. On this basis, Google maintained that Apple’s iOS represented a clear and effective competitive constraint on Android.

Neither DG COMP nor the General Court accepted this position. Instead, both institutions upheld a market delineation that separated licensable from non-licensable Operating Systems, thereby excluding Apple’s iOS and framing the market from the perspective of smartphone manufacturers.

Crucially, this conclusion was not based on a definitive rejection of Google’s arguments, but rather on Google’s failure to produce compelling evidence to support its claims – particularly the assertion that non-licensable Operating Systems exerted a direct or indirect competitive constraint on Android (see Case T-604/18 – Google (Android), paras 102–254; especially paras 146–147 and 268).

2.2 Google was dominant in the identified market

Given these narrowly defined product markets, Google’s market share appeared exceedingly high – ranging between 70% and 90 % – leading to an inevitable finding of dominance across all markets (Case AT.40099 – Google Android, recitals 446, 596–598, and 681–685). Nevertheless, the case illustrates a more nuanced and balanced approach to market definition and dominance analysis, particularly within digital markets.

While the General Court ultimately accepted DG COMP’s market definitions, it also offered critical reflections on the limitations of a rigid market delineation approach. In particular, the Court acknowledged that digital markets are often highly interrelated, with overlapping functionalities and complex user dynamics. This interconnectivity, the Court noted (Case T-604/18 – Google (Android), paras 109 and 116–120), calls for a multi-level or multi-dimensional analysis that goes beyond traditional indicators such as market shares. Factors such as innovation, control over data, and user behavior must also be considered.

Moreover, although DG COMP rejected the inclusion of Apple’s iOS in the relevant product market, it nevertheless considered whether Apple and other vertically integrated providers exerted indirect competitive pressure on Google – a factor integrated into the dominance assessment (Case AT.40099 – Google Android, recitals 147 and 268). While this argument was ultimately not upheld, its consideration reflects an evolving methodology – one that allows for the operationalization of a multifaceted dominance analysis, where market power is not treated as a mere function of narrowly defined markets and numerical thresholds, but is instead evaluated in light of broader ecosystem dynamics.

2.3. Google’s motive to engage in the abuse

Recognizing that abuse may be defensive in nature – intended to protect an already established dominant position – is crucial for understanding the Google Android case and other abuse cases involving Big Tech. Unlike more traditional competition cases – including Google Shopping – where dominance in one market is leveraged to enter or monopolize another, the conduct in Google Android largely involved leveraging power within markets already dominated by Google. Specifically, Google was found to have used its dominance in the markets for a) licensable smart mobile Operating Systems and b) Android app stores to protect and reinforce its position in the market for c) general search services. In this instance, dominance was not used to expand into a new market but to foreclose potential threats to an existing stronghold.

Because Google already held a dominant position in the general search market, it had comparatively little to gain – but a great deal to lose – if the market evolved in a way that diverted users to rival services. This provided a strong incentive to ensure that Google Search remained the default choice across platforms. As outlined earlier, Google appears to have recognized early on the strategic importance of default status and the risks posed to its search business by shifts in user behavior and technological developments.

Additionally, the potential proliferation of Android “forks” – modified versions of the Android Operating System – posed a direct threat to Google’s control over the Android ecosystem and, by extension, the search market. Thus, efforts to restrict or regulate the development and distribution of Android forks also served a defensive purpose, aimed at safeguarding Google’s broader commercial interests.

3. An abuse with three linked elements but one objective

Once Google was established as dominant, Article 102 became applicable to assess potentially exclusionary conduct. Case law has progressively emphasized the importance of evaluating whether the conduct in question risks excluding an As-Efficient Competitor (AEC). In Google Android, the General Court confirmed the relevance of the AEC standard, reaffirming that it is for DG COMP to demonstrate a plausible risk of foreclosure (Case T-604/18 – Google (Android), paras 639–643). The Court also acknowledged DG COMP’s margin of appreciation in this context, including its discretion to take into account post-decision events and market developments to support its assessment (paras 89–90) if relevant.

Of the involved actions, only the tying abuses were accepted by the General Court. The exclusivity abuse – linked to Google’s Revenue Sharing Agreements (RSA) – was rejected, primarily due to what the Court identified as a misapprehension of the scope and impact of the payments, and consequently, their potential anti-competitive effects (Case T-604/18 – Google (Android), paras 657 and 693–698).

In its assessment of the RSA, the General Court found that DG COMP had committed several methodological errors, particularly in its evaluation of (i) the size of the contestable market share, (ii) the costs faced by an As-Efficient Competitor (AEC), and (iii) the feasibility of such a competitor matching Google’s payments (paras 798 and 962–1005). Furthermore, the Court held that DG COMP had infringed Google’s right to be heard, particularly in its handling of the AEC test’s structure and application.

As a result of these procedural and substantive deficiencies, the General Court concluded that the RSA-related payments did not constitute separate abuse under Article 102. Rather, they were found to reinforce the anti-competitive effects of the Mobile Application Distribution Agreement (MADA), without constituting an infringement in their own right (paras 451, 800–802, 1005, and 1018). Accordingly, the Court held that a partial reduction of the fine was justified – from €4.34 billion to €4.12 billion (paras 1032–1114).

4. Google’s appeal to the Court of Justice and the AG Opinion

Before the Court of Justice, Google did not contest its dominant position (AG Opinion, para 18), whereas DG COMP accepted the annulment of the finding of abuse concerning the Revenue Sharing Agreements (RSAs) (AG Opinion, para 43). The case was therefore limited to three main issues: (1) the alleged abusive tying practices, (2) the application of the doctrine of Single and Continuous Infringement (AG Opinion, para 44), and (3) the calculation of the fine. In this, Google rested its appeal on six grounds.

4.1. The MADAs had been misinterpreted

In its first and second grounds of appeal, Google challenged the General Court’s assessment (AG Opinion, para 45) of: a) the causal link between the Mobile Application Distribution Agreements (MADAs) and their alleged exclusionary effects, and b) the capacity of the MADAs to produce anti-competitive effects on an As-Efficient Competitor.

4.1.1 No Causal Link Between the MADAs and Exclusionary Effects

To support its first ground of appeal – that there was no causal link – Google raised three main arguments (AG Opinion, para 48). It claimed that the General Court erred in law by: (i) considering the RSAs as contextual elements, regardless of having been rebutted as abusive (ii) failing to distinguish between the effects of default settings and those of pre-installation, and (iii) not accounting for the state of competition that would have existed in the absence of the MADA pre-installation conditions.

Regarding the first point, AG Kokott rejected the argument (AG Opinion, paras 61–69), offering three main observations. First, factual elements may be included in the assessment even if they are not themselves abusive. Second, when an infringement is characterized as a Single and Continuous Infringement, it may comprise elements that are not independently abusive. Third, while a counterfactual analysis may sometimes be useful, it was not meaningful in this case. Google had made access to the RSAs contingent on the signing of the MADA, which in turn was necessary for access to the Play Store. Moreover, the market was characterized by dynamic network effects, making it practically impossible to analyze how each contextual element, taken in isolation, specifically contributed to the exclusionary effects.

Regarding the alleged failure to distinguish between pre-installation and default settings, Google referred to how only the former had been recognized as abusive. Consequently, the General Court should not have permitted evidence concerning default settings to influence its assessment of pre-installation practices. AG Kokott considered this argument largely inadmissible, as it effectively called for a reassessment of the facts (AG Opinion, paras 78–84). Insofar as the argument was admissible, it could be refuted on the same grounds previously used to dismiss the first claim concerning an alleged failure to properly assess the effects. Moreover, in practice stakeholders tended to conflate the two concepts.

Concerning Google’s argument about the lack of a counterfactual analysis, AG Kokott again found no merits (AG Opinion, paras 92–97). She emphasized that the assumption that a case can only succeed if based on a counterfactual analysis was incorrect. DG COMP enjoys a margin of discretion, and the pre-installation conditions in the MADAs provided Google with a competitive advantage that rivals could not offset. Furthermore, while counterfactual analysis may be informative, it is only useful when both meaningful and feasible.

4.1.2 Absence of Exclusionary Effect

In support of its second ground of appeal, that the MADAs could not produce anti-competitive effects, Google pointed to the ability of end-users to download alternative applications, thus allegedly neutralizing any advantage conferred by pre-installation.

AG Kokott (AG Opinion, paras 106–108) acknowledged that EU case law adopts an effects-based approach to tying practices. She then find that the General Court had correctly applied this standard, (AG Opinion, paras 125–145 and 146-157), including its assessment of whether competing applications were pre-installed by manufacturers prior to the sale of mobile devices or whether their subsequent installation by end-users was realistically feasible. This reasoning applies generally, and specifically in relation to the tying of the Play Store with Google Search and Google Chrome. Consequently, AG Kokott recommended rejecting Google’s argument.

In this process, AG Kokott (AG Opinion, paras 128-140), addressed the relevance of the As-Efficient Competitor (AEC) test, noting how companies remained at liberty to submit such, and how it holds value for price-based and non-price-based abuse. Given the characteristics of the digital sector and Google’s exceptionally high market shares, the AEC test was not appropriate or necessary in this context.

4.2. The AFA had not been properly evaluated

In its third and fourth grounds of appeal (AG Opinion, para 159-160), Google challenged the General Court’s findings on three fronts, by arguing that the Court had a) rewritten DG COMP’s conclusions regarding the Anti-Fragmentation Agreements (AFAs); b) attributed the alleged exclusionary effects of the AFA to conduct that had not been found abusive, and c) not adequately consider the AFAs as being objectively justified.

In DG COMP’s Decision, DG COMP had expressed concerns about how the AFAs were implemented, while at the same time acknowledging that they pursued a legitimate aim. The General Court clarified (AG Opinion, para 165) that the AFAs were deemed abusive only to the extent that they restricted manufacturers from selling devices running non-compatible Android forks – even in cases where no Google applications were pre-installed. However, Google argued that this interpretation diverged from DG COMP’s original findings, effectively accusing the General Court of having rewritten DG COMP’s Decision. This submission was ultimately unpersuasive; AG Kokott firmly rejected the argument (AG Opinion, paras 170–180), finding no indication that the Court had distorted or substituted DG COMP’s reasoning.

As to the second aspect of the third ground of appeal, Google contended that the AFAs did not hinder the development or use of incompatible Android forks, since manufacturers were generally unwilling to market such devices. Google maintained that a proper counterfactual analysis would have revealed this, thereby undermining the General Court’s conclusions. Once again, AG Kokott disagreed (AG Opinion, paras 186–188), emphasizing that this line of argument relied on the erroneous assumption that counterfactual analysis is a prerequisite for establishing abuse. Furthermore, she emphasized Google’s strategic interest in restricting the proliferation of Android forks, as such forks could compromise Google’s control over the broader Android ecosystem – an interest acknowledged in the original assessment.

Regarding Google’s assertion that the AFAs were objectively justified, the fourth ground of appeal, Google argued that the General Court failed to evaluate this defense adequately. AG Kokott (AG Opinion, paras 191–200) recommended dismissing this claim, primarily on grounds of inadmissibility, as it amounted to a request for a reassessment of the facts. Moreover, the claim was predicated on the same flawed insistence on counterfactual analysis previously rejected by the AG. It also overlooked the fact that the General Court had, in fact, addressed the issue of justification in its judgment.

4.3. The Single and Continuous Infringement and the fine

In its fifth ground of appeal, Google argued that the General Court erred (AG Opinion, para 202) in accepting the inclusion of the Revenue Sharing Agreement (RSA) payments within the framework of a Single and Continuous Infringement, on the grounds that these agreements did not, in themselves, constitute abusive conduct. However, in the view of AG Kokott (AG Opinion, paras 209–215), this argument was ultimately immaterial. The remaining instances of abuse had been upheld, and the RSA pursued the same overarching anti-competitive objectives as the other forms of misconduct. Accordingly, there was no issue with the General Court’s decision to incorporate the RSA within the broader finding of a Single and Continuous Infringement.

In its final and sixth ground of appeal, Google challenged the calculation of the fine (AG Opinion, para 217). However, AG Kokott (AG Opinion, paras 225–241) found no merit in this argument and recommended its dismissal.

5. Google Android in reflection

The General Court’s ruling in Google Android was well-reasoned and should be broadly welcomed by the antitrust community. The ruling reaffirms the applicability of Article 102 in addressing abusive leveraging by dominant digital platforms, whether such conduct targets adjacent markets, as in Google Shopping, or serves to ring-fence an existing ecosystem, as in Google Android. Importantly, the judgment underscores that intervention under Article 102 is appropriate only when the evidentiary record plausibly supports a finding of abuse, thereby mitigating the risk of over-enforcement. The guiding standard remains the holistic assessment of “all relevant circumstances,” and in particular, whether foreclosure is plausible – an approach set out in Intel and reaffirmed in Google Android.

In this context, it is significant that AG Kokott not only recommended that the judgment be upheld in its entirety but also delivered a well-reasoned and meticulously drafted Opinion. Her Opinion contains several bold, yet factually inaccurate, assertions. For instance, when opening by claiming how Google search engine is omnipresent in the daily lives of the majority of the world’s population (AG Opinion, para. 2). This assertion overlooks significant geopolitical realities: Google has limited operations in China and is entirely blocked in Russia – two countries that together account for approximately one-fifth of the global population.

Additionally, AG Kokott references older cases such as Tetra Pak and Hilti (AG Opinion, para. 106) as foundational precedents for identifying tying as abusive conduct suggesting this as the base line. This reliance appears misaligned with recent case law, her own observations, and DG COMP Article 102 enforcement framework, as articulated in its guidance papers.

Despite these problematic statements, AG Kokott’s analysis is otherwise rigorous. She systematically examined and rebutted Google’s arguments, even those directed at reevaluating factual matters, which are in principle inadmissible at this stage of judicial review. In evaluating potentially abusive conduct, DG COMP retains a degree of discretion, provided its analysis is coherent, methodologically sound, and based on reliable and transparent data. AG Kokott affirms this principle and underscores the relevance, where feasible, of analytical tools such as the As-Efficient Competitor (AEC) test and counterfactual analysis. She even confirmed the availability of the AEC test for price-based and non-price-based abuse.

The next step will be the actual judgment, which will close another chapter in the Google saga.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

As I did last year through The Power of No series, I will be covering this year’s compliance workshops held by the European Commission, where the gatekeeper representatives meet stakeholders to discuss their compliance solutions to the DMA’s obligations (and the updates they introduced since 2024). Yesterday, I already covered Microsoft’s compliance workshop in detail, whereas this blog post considers Amazon’s compliance workshop and the few differences that I could spot from its previous workshop held in 2024.

Regulatory dialogue on data portability and access

In this second round of compliance workshops, the European Commission (EC) has decided to provide a bit of an overview of the points of discussion that it has tabled on its regulatory dialogue with the gatekeepers.

For Amazon’s case, the EC has taken issue with its data-related obligations connected to business and end user access to data in different forms, notably under the premise of portability under Article 6(9) and the instances of mandated transparency set out in Articles 5(9), 5(10) and 6(8) of the DMA. Amazon’s compliance approach with these provisions was exhaustive and detailed, as stemming from its 2024 compliance report (and from the details it presented on the previous iteration of the compliance workshop last year). The vast majority of the technical solutions it already put forward in 2024 remain in place, such as the APIs accessible to third parties to exercise portability of end user data or the Seller API, allowing sellers and authorised third parties by them to integrate tools to receive access to data on their business operations within the Amazon marketplace. Amazon’s legal representatives went over the advantages and features of those once again this year.

Nonetheless, the impending issues surrounding Amazon’s compliance approach highlighted by the EC went back to broader notions of principle. For instance, the EC takes issue with the complex validation process that Amazon created to provide authorised third parties the possibility to connect to the data portability APIs. In the regulator’s view, Amazon established too many restrictions to enable third-party access to sensitive data, classified by the gatekeeper as Category 2 data, such as the end user’s shopping history, shopping wishlists, interest-based ads preferences, or contact details. One cannot say that Amazon introduced these limitations without any justification, insofar as its legal representatives highlighted at the workshop that over 75% of the applications they received for such data types corresponded to non-EU-based applications from data aggregators, who are mainly based in countries with EU adequacy decisions pursuant to the GDPR. In other words, business users or third parties that cannot, generally, access any type of personal data from data subjects located in the EU. In any case, the gatekeeper accommodates some of the EC’s petitions by, for instance, streamlining some of the questions that Amazon made to third parties to screen them on the validation process for accessing the portability API (as already set out on page 54 of their recent 2025 compliance report). Amazon’s representatives stressed the need for further guidance to be issued by the EC on these aspects overlapping with the data protection framework, hopefully through the release of guidance alongside the EDPB, which both institutions ensure is currently underway.

Alternatively, Amazon also established additional controls for end users to be able to consult the data authorisation’s status, especially in those cases where such action has expired or been cancelled directly by the customer (page 45 of the report). In a similar vein, Amazon also goes to great lengths to accommodate more transparency into the process for business users to measure their performance within the gatekeeper’s CPS pursuant to Article 6(10) by, for instance, enhancing a new dedicated ‘Amazon Seller Data Access’ help page to increase awareness on its data solutions (page 75 of the report).

In its initial remarks, the European Commission also briefly referred to its ongoing discussions with the gatekeeper relating to the ad transparency provisions. As pointed out by the regulator, there are some areas of Amazon’s compliance that are being questioned and which are currently being analysed, such as the modalities of access to the data or data granularity. Although Amazon’s legal representatives did review the gatekeeper’s compliance solutions in detail for these provisions, their regulatory position has not changed since 2024, since the data on the ground shows that business users (including advertisers and publishers) seem reasonably satisfied with the available APIs and processes available to them.

Price gouging and the DMA’s harmonisation objectives

On the first set of interventions to the workshop, Amazon delivered some provocative remarks, building on former VP and Commissioner Vestager’s words when the DMA was being discussed in the legislative process; “we want a single European rulebook. We have tabled this proposal to avoid fragmentation”.

Even though the DMA’s approval and entering into force entailed that some legislative proposals by the French and Dutch administrations aimed at establishing digital rules addressing market power were stopped on their own tracks, one of the Member States did go under the radar: Germany. Through the approval of Section 19a to its competition law regime (GWB), the German competition authority designated undertakings with a paramount significance for competition across markets, Amazon being one of them. The debate that ensued before the Courts made just the more evident the collision between the DMA as a European regulation and the potential application of the principle of primacy to Section 19a GWB. The Bundeskartellamt had enforced Section 19a GWB in a few instances (e.g., on Google’s data processing practices, see here). A few weeks ago, however, the German competition authority issued an SO against Amazon relating to its price control mechanisms and the way by which it reviews sellers’ prices to altogether remove some third-party seller offers from the Marketplace.

This enforcement action through the German national competition law regime runs in parallel to the EC’s own monitoring of Amazon’s compliance, considering Article 5(3) DMA, to the extent that the regulator is currently discussing with the gatekeeper on those measures that could bring an equivalent effect to parity clauses under the provision. As set out by Amazon’s legal representatives on the workshop (and on Annex I to its compliance report in pages 5-7), to ensure that offers by third-party sellers only show competitively priced products, it applies certain governance rules to protect customers from significantly high prices or from price errors/egregiously high prices. The gatekeeper considers the prices that third-party sellers set in reputable competing stores to set out the lowest price in the market as a benchmark to compare Amazon with the outside-of-Amazon pricing. Based on this yardstick, when an offering on a competing marketplace is available for less than the offer in Amazon Store, the gatekeeper will not include that product as the Featured Offer (FO), which is the prominent section on a product’s detail page where customers can easily add items to their cart or make a purchase. In this same vein, if Amazon spots that the price shown by the third-party seller is egregiously higher, it will eliminate the offer from its platform.

Amazon failed to provide further details to questions from the participants on how those prices are set and from what reputable competing stores they are extracted from. Notwithstanding, the main concern here is not only that Amazon performs such governance of its platform. Instead, the EC takes issue with the fact that Amazon, via these indirect means, may deter third-party sellers from selling on other platforms (or on their own websites) for lower prices than it does on its own platform, aka bringing an equivalent effect to those impacts produced by parity clauses. Needless to say, the Bundeskartellamt’s arguments are quite similar, but they are translated into competition terms (and not the terms of broader contestability and fairness concerns).

To this call, Amazon provided an unconvincing explanation of the reasons justifying its imposition of this lowest-price benchmarking to show the FO. Its legal representatives simply stated that the system does not benchmark seller prices on Amazon vis-à-vis their prices on other platforms, and it does not restrict sellers from pricing lower elsewhere. Amazon believes that the European Commission already reviewed the system under Article 102 TFEU through its Amazon Buy Box case closed in 2022 with commitments. As a matter of fact, if one navigates to the third commitment presented by Amazon on that case it already established that for setting the FO it “may use factors that are objectively justified in order to protect consumers from the risk of Seller fraud and abuse when deciding whether a Seller qualifies for participation (…)”. In my own mind, two sets of cases must be set apart in terms of considering whether Amazon complies with Article 5(3) DMA, as shown in the table below.

From the outset, it seems that Amazon deliberately conflates the first restriction’s impacts on the FO under the justification of the second type of limitation to automatically assign them into the Amazon Buy Box’s scope. But that does not mean that it should. Under the premises of proportionality and necessity, one should not follow from the other. Having said that, and regardless how unconvincing Amazon’s scarce statements on the workshop were, from Amazon’s compliance report (Annex I, pages 6-7) we learn that the EC has already intervened by requesting information on the composite benchmark value used for both tools and they have confirmed (according to the gatekeeper, that is), “that Amazon does not require Sellers to change their price in the Amazon Store to match their off-Amazon pricing on third-party online intermediation services or own direct to consumer websites”. The matter is now whether this conclusion is credible or not, bearing in mind that stakeholders participating in the workshop already highlighted that, in practice, third-party sellers respond to the application of this system by catering to different product types on Amazon as opposed to those provided on different marketplaces.

Same ol’ same ol’ self-preferencing

As the renowned meme reads, here we go again. Back on Amazon’s first compliance workshop, its allegations that it did not self-preference its Amazon Retail products vis-à-vis third-party seller products through its ranking algorithms did not qualify as unconvincing. Nor did they on this second iteration of the workshop, because we heard the same arguments on repeat.

According to Amazon’s legal representatives, its ranking of the product research results page and the product detail page remains neutral and non-discriminatory because it relies on objective inputs that are transparent, fair, and non-discriminatory. Those same tenets not only apply to its general ranking of results for a particular user query, but also to the rest of its features, such as Sponsored Ads or Widgets. The ranking criteria used, for instance, to display ‘trending now’ offerings are not based on the seller catering to them or who provides the logistics for their delivery, but rather on objective criteria such as an offering’s popularity. Asked on how it performs continuous monitoring on the application of these objective criteria on its ranking, Amazon highlighted it had created a forward-looking mechanism to ensure that any changes to the ranking inputs and process are properly reviewed by an Amazon team so as to ensure they are DMA compliant. A vague response to a vague premise, but the European Commission has not yet challenged through its enforcement action, despite that it has had access to Amazon’s ranking algorithms, as confirmed at the workshop.

Call me, beep me: Amazon’s Rufus AI tool

Those who lived (or endured) the Disney Channel era will remember Rufus, the mole-rat of Ron Stoppable, creating all sorts of mayhem in every episode of Kim Possible. Amazon has its very own Rufus, not a pet, but rather a genAI-powered shopping assistant to help customers on their shopping experience. The chatbot was rolled out into some Member States in the EU (France, Germany, Italy, and Spain) through its beta version late last year and has caused some confusion ever since.

Like any other AI system, Rufus suffers from hallucinations. For instance, if asked for the cheapest options of a given product, it simply does not deliver them. Amazon acknowledged that Rufus could spew some dysfunction when responding to user prompts since “it’s still early days for generative AI, and the technology won’t always get it exactly right”. Notwithstanding, Amazon’s legal representatives extolled Rufus’ virtues by demonstrating the capabilities it would bring to the market and to Amazon Store in particular, e.g., comparing different options relating to the same product type, such as a drip vs. a pour-over coffee maker. In this sense, they did indicate that Rufus is an integral part of the Amazon Store and that it is DMA-compliant. The gatekeeper highlighted that Rufus makes decisions to respond to user prompts regardless of the seller (whether Amazon Retail or a third-party seller) and those in charge of performing the delivery and logistics of the product (page 21 of Annex I of the compliance report). In other words, Rufus aligns with the regulatory requirements of Article 6(5) DMA.

To questions of the workshop participants on its potential misalignment with Articles 5(2) and 6(2) DMA, Amazon’s legal representatives did not provide any immediate or satisfactory reaction. For instance, when asked about how the processing of personal data for training Rufus (and also for user-inserted data) worked, they simple reiterated some of the previously available information on the tool; that Rufus is trained on publicly available data from across the web and on data from its Amazon Store such as its product catalogue, customer reviews and community Q&As, so that the legal requirements under Article 5(2) DMA did not activate as a consequence. It is unclear whether ‘publicly available data’ is short for web scraping. In that particular case, Amazon would not result to be unscathed from the application of Article 5(2) DMA, as I point out in a recent paper. On this same note, the gatekeeper’s legal representatives did confirm that Amazon was considering rolling out ads on the AI tool, which would make cross-using and combining data across its two CPSs unavoidable. In the end, Rufus may (once again) cause a headache or two to the main characters of the DMA story.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

As I did last year through The Power of No series, I will be covering this year’s compliance workshops held by the European Commission, where the gatekeeper representatives meet stakeholders to discuss their compliance solutions to the DMA’s obligations (and the updates they introduced since 2024). This blog post covers the first workshop organised by the EC in 2025, targeting Microsoft’s technical implementation of the regulation.

Regulatory dialogue with the European Commission

As opposed to last year’s compliance workshops, the EC was a bit more open in terms of disclosing some of the terms and content that it is currently discussing with the gatekeepers. At the start of the workshop, EC officials quickly went through the main points of dialogue that it is currently discussing with Microsoft regarding its compliance with the DMA. On that note, the EC highlighted its focus on Microsoft’s implementation of Articles 5(2), 6(9), 6(10), 6(3), and 6(7) DMA. In other words, these are the EC’s regulatory priorities to take Microsoft to an effective enforcement scenario.

For instance, the regulator stressed its focus on Microsoft’s introduction of new AI features on its two core platform services (CPSs), Windows PC OS and LinkedIn, insofar as those could be at odds with some of the regulation’s prohibitions, notably that of combining and cross-using personal data across services as set out in Article 5(2) DMA. Microsoft went into detail throughout the workshop to try to satisfy the EC’s thirst for covering all the AI grounds that have not been formally addressed by any other piece of EU regulation.

In this same sense, the EC underscored one of its top priorities in terms of the data portability solution (aka Article 6(9) DMA): Microsoft’s future release of Recall, a functionality termed by the gatekeeper as the ‘photographic memory for Windows’ integrated into Copilot+ PCs running on Windows 11. That is to say, Recall will not be available on every single device running on Windows PC OS, but only on Copilot+ PCs, which are specifically designed for performing AI tasks with additional computing power that outpaces Intel and AMD laptops. The Recall feature takes images of the user’s active screen every few seconds to record (or recall) all its past activities, including sensitive data and passwords. Upon the live recording of one’s screen, the user will then be able to transcribe and translate video meetings or search what they were viewing at a particular point in time by scrolling through a timeline of snapshots. Recall is already available in other jurisdictions of the world, such as the US, although it was forced to hold back its release after security concerns were raised ahead of its debut in June 2024. Some commentators even labelled the functionality as spyware, although Microsoft defended that Recall only stores data locally on the user’s PC and is protected by a PIN and not in the cloud. Due to the backlash, Microsoft also made Recall uninstallable on Copilot+ PCs and transformed it into an opt-in experience for the users to choose whether they wanted to enjoy the (surveillance) experience. The feature will not be available in the EU until later this year, as some reports confirm.

Surprisingly, the EC declared it was not analysing the Recall feature in the context of the data-related obligations of Articles 5(2) or 6(2) DMA. Instead, it is seeking feedback on how data available to the Recall feature may be instrumentalised by third-party business users under the data portability solutions. The workshop participants also remained moot on the soon-to-be-released future, although its rollout may provoke tensions with the encryption requirements set out, for instance, in Article 7 DMA relating to horizontal interoperability relating to messaging services such as WhatsApp and Facebook Messenger. As some critics have already argued, the Recall feature undermines the security of encrypted apps like WhatsApp and Signal by storing anything shown on the user’s screen, including private messages. Additionally, cybersecurity experts have already demonstrated that guessing the PIN gives full access to all screen content (deleted or not), including sensitive conversations, images, and passwords. Once more, the DMA is set at a crossroads with data protection standards and cybersecurity risks to be factored into the regulatory mix.

Reporting on progress: data portability, access, and combinations

After the EC’s intervention setting out the stage for a convivial discussion between the gatekeeper and stakeholders, the gatekeeper went into the thick of it by presenting the changes that it had made since March 2024 relating to its compliance with some of the data-related obligations on its LinkedIn service, notably Articles 5(2), 6(9) and 6(10).

To demonstrate compliance with the prohibition of combining and cross-using personal data across its services, Microsoft representatives brought forward its LinkedIn consent moment that it had already included within its 2024 compliance report (pages 3-12 of the report). In line with the consent experience, LinkedIn informs users about the way in which it connects services and types of data to deliver its core services. According to Microsoft’s legal representatives, the DMA consent moment is particularly targeted at informing users about the different services it caters to them, i.e., LinkedIn, LinkedIn Jobs, LinkedIn Marketing Solutions, and LinkedIn Learning, since combinations of personal data can only occur based on the premise of the consumer’s consent. To this effect, stakeholders asked about the user adoption rates relating to the granting of consent and Microsoft’s representatives confirmed that a meaningful portion of EEA members agreed to the consent moments, but they also had seen non-minimal responses from a portion of LinkedIn members not granting consent or minimising the volumes of data they granted access to. Despite the vehemency of some stakeholders in their questions addressed to Microsoft relating to its testing of the consent moment, the gatekeeper established that it had conducted no testing on the prompt’s language neutrality.

Some participants in the workshop also tried to test the waters resulting from the Cologne Higher Regional Court’s ruling on the granting of interim measures, where it interpreted Article 5(2) DMA in the context of metering whether Meta’s integration of data from its Facebook and Instagram services to its AI models. According to the Court, there was no fundamental tension with the DMA provision since there is no ‘merging’ of data because the gatekeeper does not combine data from user profiles on different services or from other sources with regard to a single, specific user. Participants in the workshop asked (by dropping references to Meta) whether Microsoft also considered that data combinations should apply for the same user, i.e., user profiles to be captured under Article 5(2) DMA. Microsoft’s representatives did not deviate from their main line of argument but inadvertently confirmed that the prohibition on data combinations applies across end users, regardless of whether a user profile may be provided therein.

Furthermore, Microsoft introduced the GDPR consent moment that it launched in December 2024 to satisfy the Irish data protection authority’s concerns in its processing of personal data under the data protection regulation, as set out in pages 3-10 of its 2025 compliance report. Both consent prompts overlap substantially in terms of the data they refer to. Microsoft settled that it honors user choices made across the two of them in those cases where conflicting decisions relate to the same group of data. Additionally, stakeholders called out the gatekeeper on its implementation of Article 5(2)(d), compelling it to not sign in end users to other services of the gatekeeper to combine personal data, to the extent that Windows PC OS users are automatically signed in to its cloud storage service OneDrive upon the installation of Windows 10 and 11 with a Microsoft account. Microsoft’s legal representative defended the gatekeepers’ practice, insofar as the automatic sign-in to OneDrive was not performed for the purpose of combining data, but rather to provide easier access to online storage, as it already set out on pages 19-20 of its 2024 compliance report.

Moving forward on the data front, Microsoft did not report any additional changes to its compliance approach regarding Articles 6(9) and 6(10), except for the few instances where it reiterated its promises of continuous improvement of its LinkedIn Member Data Portability API, in terms of the features available to business users. On this front, the gatekeeper did provide some tangible evidence of its progress on the implementation of both provisions, since as of last year, 921,000 API calls were completed successfully stemming from the API program designed for LinkedIn and more than 2 million API calls were conducted relating to those APIs aimed at granting access to business user data pursuant to Article 6(10) DMA. Stakeholders voiced their concern about the low response rates of those same APIs and the limitations imposed on business users relating to the number of calls they could direct at the same API. Microsoft legal representatives justified the initial problems with the rolling out the APIs due to the sheer volume and complexity of the data involved when it comes to the portability of its LinkedIn users’ data. Nonetheless, they assumed the compromise to investigate undue delays in those response rates or delays caused by technical issues, as well as to increase call limits based on internal reviews conducted as a consequence of business user requests.

Default settings, interoperability, and an evolving compliance strategy

In the second half of the workshop, Microsoft went on to describe how it constitutes an open platform for developers.

In line with the requirements set out under Article 6(4), users can install an application or app store onto Windows PC OS from any source without payment or permission from it. Predominantly, those downloads do not happen on the Microsoft Store, the gatekeeper’s own app store for its operating system, but on the Internet. In this respect, the gatekeeper’s business model revolves around the third-party developers’ control over the technical and commercial experience within their applications and application stores.

Having said that, Microsoft presented to the workshop assistants the changes it had recently introduced to comply with Article 6(3) DMA to the requisite legal standard that the EC had set out for it to comply with, as stemming from their informal exchanges. Most of those changes reproduced the announcement Microsoft had already made on the 2^nd of June on its blog. During intense debate, Microsoft defended that the few changes it introduced this year to its DMA compliance report do not point to the idea that it did not comply with the regulation from the first compliance deadline, i.e., March 2024. Instead, the gatekeeper upholds its willingness to engage with the EC’s feedback and accommodate its points of view to the greatest extent possible, bearing in mind its broader business model.

Following its engagement with the EC, Microsoft decided to make its Microsoft Store uninstallable, imitating Apple’s example, whereby it promised to make its App Store uninstallable for all users. Similarly, Microsoft also decided to budge in terms of the way users can change defaults on Windows PC OS, specifically to address the unique situation of file and link types. In other words, the EC pressured the gatekeeper to make it easier for users to change their defaults not only directly on an application (e.g., a browser) but also through settings depending on the type of file concerned (e.g., a PDF or a JPG file).

To that end, Microsoft will start rolling out in July 2025 functionality to: i) ensure that users can set any given browser as the default for all file types, except for pdf files; ii) automatically pin the user’s chosen default browser on the Taskbar and Start of its operating system, unless the user decides to unpin them explicitly; and iii) provide a separate control for pdf controller to set defaults for this file type via a one-click set default button. Stakeholders called for further action in this respect and asked whether Microsoft would consider re-designing its default settings to accommodate the establishment of browser defaults only through one click. The gatekeeper’s reaction was slightly discouraging since Microsoft’s view is that users should be able to set defaults through Settings so that they remain in control of every single change they perform on their devices.

Alternatively, Microsoft will cease prompting users to make its proprietary browser Edge a default when it is not opened on the user’s screen by May 2025. Aside from satisfying the EC’s expectations in unbundling all digital services in the broadest sense possible, these few changes may also contribute to appeasing the calls for the reversal of the non-designation decision issued by the EC of its Edge service, which is currently under appeal under pending Case T-357/24.

Notwithstanding, some participants of the workshop still questioned Microsoft on the default settings ingrained into S Mode, a version of Windows 11 designed for security and performance. According to the S Mode requirements, users must move out of S Mode completely (and onto Windows 11) to take advantage of the change of default settings from Bing and Edge and thus, the DMA provisions do not apply to those laptops operating on this particular Windows 11 version. Microsoft’s legal representatives clarified this aspect of their compliance approach by setting out that under 10% of the machines in the EEA operate on S Mode, and more than 80% of them switch out of S Mode. On one side, they argued, that if 80% of users manage to switch out of the version, that demonstrates that it is not unduly difficult to do so. On the other side, a different configuration of S Mode would undermine its ultimate purpose, which is to provide a Windows PC OS version directed at novice users with small leeway for its configurability on the end user side.

Workshop participants also circled back to Microsoft’s compliance report, in particular to page 82, to question specific carve-outs to the default rule. Specifically, the compliance report states that “some applications are designed to use a specific application that is not the default application because it may provide the user with a more consistent end-to-end experience or enable the use of new features that are supported by the chosen specific application”. Legal representatives clarified that only two Microsoft applications rely on its browser Edge, Microsoft Teams, and Outlook. Both of those have controls where the user will be prompted on whether to choose the default browser to view a particular link or to use Edge, since the latter provides innovative technology to, for instance, display Outlook side-by-side with Edge. In their own view, such a situation does not breach Article 6(3) DMA because users remain in control of what web browser to choose when clicking on a particular link.

In a similar way, Microsoft’s technical implementation of the DMA also evolved with respect to its interpretation of the interoperability solution under Article 6(7) DMA. At the previous compliance workshop held in 2024, the gatekeeper unbundled the Taskbar on Windows 10 from its Bing search engine and its feed extensibility of the Windows 11 Widgets Board, formerly backed by Edge’s browsing experience.

Taking a step further, Microsoft has fundamentally re-worked its approach to providing access to business users to its Windows Search (desktop functionality enabling the finding of files, apps, settings and web information) and Widgets Board (feature providing a collection of interactive elements providing quick access to information and actions from various apps and services) functionality. For Windows Search, it will enable third-party web search providers to power it and will operate a redesign of the functionality to ensure they are automatically enabled to do so upon their installation. Those changes, according to Microsoft’s representatives, will be rolled out in June 2025. Moreover, the gatekeeper has eliminated restrictions on access to the Windows Search functionality, whereby only applications available on the Microsoft Store could be shown on it. Since May 2024, it has eliminated the main requirement for its shipping. Additionally, when users click links on Windows Search or in Widgets Board, they will no longer be opened on Bing (nor will they be prompted to reinstall Edge). Once the change is finally rolled out in July, the user’s default browser will be opened as a consequence of the user’s interaction with the gatekeeper’s proprietary functionality on the desktop.

GenAI tools intersecting with the DMA provisions

If there is a buzzword that has been making the rounds relating to the DMA’s enforcement, it is generative AI. I have discussed the matter extensively in a couple of papers (see here and here). And the topic of AI has flooded the workshops’ agendas, not only for Microsoft but across all gatekeepers.

Microsoft’s approach to metering the intersection between AI functionalities and the DMA has been quite distinct as opposed to the rest of the gatekeepers. In its 2025 compliance report, Microsoft already declared how it resolved the fundamental tension between Article 5(2) DMA and the generative AI-reliant tools it had introduced on its LinkedIn platform (pages 13-14), whereas other gatekeepers have preferred not to make any written disclosures to that effect. In this context, Microsoft already declared that to train its large language models (LLMs) to produce content and enhance its LinkedIn services, it had taken two immediate courses of action. First, it does not use EEA member data to train LLMs to generate content. Second, even in those cases where it was to do so, it would honour the EEA LinkedIn member’s DMA consent settings for relevant data combinations. For instance, if an end user did not provide consent for combining personal data at the consent moment, that same data would lack a legal basis to be inputted into the LLM.

Although workshop participants tried to corner Microsoft into stating that it relied on the legitimate interest basis under Article 6(1)(f) GDPR to process personal data for the purposes of training and input data on its LLM, LinkedIn stood strong in defending that it did not use EEA member personal data for that purpose, so that it was not compelled to rely on any legal basis for justifying the processing of its personal data.

Despite a few references to AI made it to the Windows PC OS Annex to Microsoft’s compliance report, the gatekeeper was eager to demonstrate how it would integrate AI solutions on Windows PC OS and how those would be regarded from the DMA perspective.

First, the gatekeeper’s legal representatives reiterated that Windows operates as an open platform designed to deliver innovative and new applications. This same philosophy will apply to all AI applications running on Windows. Along these lines, it presented the two applications it runs on Copilot, its proprietary generative AI chatbot: Microsoft Copilot (for consumer scenarios) and Microsoft M365 Copilot (for work environments). Those applications run on Windows through public APIs that are available to any given developer, relying on its cloud infrastructure. In DMA terms, Microsoft has ensured that those APIs are available to all developers to build an app running on Windows. No further reference was made, however, to the dependency of these applications run by Windows with its own cloud infrastructure and cloud computing services, Microsoft Azure. It is yet unclear whether business users will be able to choose what cloud computing provider they choose to power their apps, even if they rely on Copilot, or if there is a technical bundle between both offerings.

Furthermore, Microsoft highlighted the creation of a new configurable keyboard key (the Copilot key) to make it easier for its users to launch an AI application when interacting with its Windows PC OS (see Image 1 below for reference).

According to the gatekeeper, when a user presses such a key, the operating system will redirect it to Copilot or, alternatively, to a competing AI service, as it already set out on page 123 of its compliance report. As shown in the screenshot above, the Copilot key currently shows the Copilot logo only. Asked by stakeholders, Microsoft reiterated that device manufacturers building PCs in the past have a long practice of including application-specific keyboard keys on their keyboards and, therefore, there would be no issue with the inclusion of a different logo on the key, notwithstanding the contractual requirements and specific restrictions that Microsoft might impose on such device manufacturers.

Second, Microsoft also put forward the platform APIs it offers to better build AI experiences on Windows. Most AI applications generally rely on AI models running on the cloud. However, the gatekeeper strives to establish a hybrid approach toward computing by building feasible solutions for AI experiences to run locally on the PC. As opposed to the dependence on the cloud infrastructure, running AI locally would keep data on-device, so that the exposure to data breaches would be reduced. Lower latency would be offered, therefore, ensuring faster and more reliable interactions with AI models. Correspondingly, Microsoft already includes on its Copilot+ PCs neural processing units designed to run AI models locally in an efficient way. On the side of expanding business user opportunities, it will additionally offer three key ways to run AI models locally by providing access to new APIs so that they can access the same type of hardware as Microsoft does. For instance, Microsoft is planning to release Foundry Local, enabling the local execution of LLMs directly on a Windows device via the download and customisation of open-source LLMs from an online catalogue on demand. In addition, it has already launched several local AI models part of Windows on Copilot+ PCs to process language and images, accessible to business users through publicly documented APIs. This provides the possibility for business users to enable AI capabilities without the need to find, run, or optimise their own proprietary model. To document the real-life results of these publicly documented APIs, Microsoft included a clear example to account for developer opportunities generated within its ecosystem.

For this particular example, Microsoft shows how its proprietary Microsoft Paint running on a Copilot+ PC uses a custom model running on a neural processing unit. Specifications and documentation to access those same hardware features are made available to all developers via APIs.

In Microsoft’s view, all these efforts demonstrate its readiness to comply by default and design. Nonetheless, it is yet unclear how those will work in terms of their bundling with other services and their reliance on its proprietary cloud infrastructure (which could be designated as a CPS at any given time by the EC through the qualitative process under Article 3(8) DMA).

Finally, Microsoft also set forth the features it will incorporate into Windows, relying on AI to deliver better functionality on its operating system, including examples such as Recall or Click to Do. Once again, it seems that the gatekeeper wants to provide the impression that it advances in the market in line with state-of-the-art technology, but that’s simply not what the DMA is about. The regulation’s devil is in the details, and the AI race is no exception to it.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

Payment and Electronic Money Institutions are entities licensed to operate as a payment institutions or electronic money institutions pursuant to the Law No. 6493 on Payment and Securities Settlement Systems, Payment Services and Electronic Money Institutions (“Law No. 6493”) and the Regulation on Payment Services and Electronic Money Issuance and Payment Service Providers (“Payment Services Regulation”). Although the Law No. 6493 covers several types of payment services, the focus of this blog post is on the service that these entities provide as a payment facilitator (“PF”).

The Turkish Competition Board (the “Board”) considers PFs and banks to be competitors in the merchant acquiring market. While competing in this market renders the relationship between PFs and banks horizontal, the fact that PFs need access to the point-of-sale (“POS”) services provided by banks in order to operate in this market gives the relationship a vertical dimension as well. Hence, PFs are dependent on banks’ input to compete with banks in the merchant acquiring market. Therefore, the application of Law No. 4054 on the Protection of Competition (“Law No. 4054”) to the relationship between PFs and banks, as well as to the unilateral conduct of banks in relation to PFs, requires further consideration.

Before examining how PFs operate in more detail, it is useful to consider how the credit card market functions. The most straightforward scenario involves three parties: (i) the bank that issues and acquires the card; (ii) the merchant; and (iii) the consumer.

As shown above, the consumer (T) uses a credit card issued by bank (A) to make a purchase from merchant (M) who has a POS of bank (A). Since merchant (M) has a merchant agreement with bank (A), it has a POS of bank (A) and can accept payments made by cards issued by bank (A). Consumer T pays bank (A) the price of the purchased product, and Bank (A) pays merchant (M) the remaining of the price after taking its commission as per the agreement between merchant (M) and bank (A).

The second scenario is quite similar to the first scenario. The only difference is that the issuer and the acquirer banks are different. Therefore, two banks are involved in the second scenario.

As shown above, another bank intervenes to perform the clearing function in the second scenario. In this case, the consumer (T) uses the card issued by bank (B) to shop with merchant (M), who either does not have a merchant agreement with bank (B), or has one that is not economically viable. In addition to the above scenario, Bank B is also involved in the transaction. When T uses a credit card issued by Bank B to shop at the merchant’s store, the merchant may process the payment through Bank A’s POS. Banks A and B then proceed with the clearing and settlement process (assuming both banks are part of a settlement scheme such as BKM – Interbank Card Center).

Both scenarios involve only banks. However, a PF can be involved in both the three-party and four-party scenarios. For a better understanding of the structure, we will include a PF in a four-party scheme. So, the third scenario is the case where a PF is involved in the second scenario.

As shown, merchant (M) receives a POS service from a PF. The PF may facilitate the acceptance of payments from merchants through agreements with banks.

The way PFs work is simple: A PF enters into agreements with different banks, integrates their POS services and provides an integrated POS solution to merchants. In other words, by contracting with various banks, a PF obtains access to these banks’ POS infrastructures and offers a catalysed service to merchants. It does so by creating its own physical or virtual POS system and enabling merchants to accept card payments from multiple banks through a single integrated interface. This means that, instead of negotiating and contracting with each bank individually and having separate POS devices, merchants can meet their POS needs by entering into an agreement with a single PF in a much simpler way. However, most PFs do not have an acquirer licence. Technically, this means that they cannot create their own virtual or physical POS devices. Nevertheless, what they are essentially doing is integrating the POS of different banks. PFs are permitted to do this without an acquirer licence.

When a merchant makes a sale, the card is directed to the PF’s service, which provides an integrated POS service, and the PF directs the card to the POS of the bank offering the most favourable commission. When a consumer makes a payment to a merchant, the PF can direct one-time (single) payments through the BKM clearing system to the bank that offers the most favourable price (this can be any bank within the BKM system). In other words, neither the merchant nor the PFs need a POS agreement with the bank that issued the card for one-time payments. Therefore, the BKM’s clearing and settlement arrangement facilitates inter-bank competition for single payments.

However, this does not apply to instalment payments. In principle, in order to accept such payments, the PF needs an agreement with the bank that issued the credit card. This gives the issuing bank a monopoly on acquiring instalment payments. In other words, a payment in instalments using a card issued by Bank A cannot be acquired at a Bank B’s POS. In principle, the issuing and acquiring banks must be the same bank for a payment in instalments. The only exception is credit card families where more than one bank has an agreement to issue and acquire the same branded credit card, such as Bonus and World.

Regulatory Framework

The Central Bank of the Republic of Turkey (“CBRT”) published the Payment Services Regulation in accordance with Law No. 6493. According to Article 8 of this Regulation, all payment service providers (“PSP”), including banks, payment institutions, and e-money institutions, are obliged to make their payment account services and payment infrastructure available to other PSPs upon request. In this context, banks that own the infrastructure must open their credit card POS access to all PSPs, including payment and e-money institutions that request it. The reverse is also true. Requests cannot be refused for any reason other than security, operational or technical requirements. Furthermore, discrimination among PSPs to which this service is provided is prohibited. PSPs are obliged to provide services to all PSPs that request them under similar conditions.

Pursuant to Article 8/2 of the Payment Services Regulation, the services provided must be comprehensive enough to enable the requesting PSP to carry out its desired activities within the scope of Law No. 6493 without any problems. In other words, when sharing this infrastructure with other PSPs, the PSP must provide it in a comprehensive manner to enable other PSPs to carry out their payment services without an issue. In this context, the PSP receiving the infrastructure must maintain a seamless service in order to compete effectively. Article 8 of the Regulation obliges the PSP to inform the requesting PSP of its decision within one month at the latest. This provision has been included in the legislation to prevent the requested bank from delaying the request for any reason other than the security, operational and technical requirements set out in the first paragraph.

To sum up, Article 8 of the Regulation on Payment Services identifies the market failure in three paragraphs and provides a comprehensive solution. The CBRT acknowledged that the market for payment services can only function if all PSPs share their infrastructure with each other. The purpose of this acceptance is to create multiple alternatives for each PSP and increase competition among these alternative channels. Not only does the provision impose an obligation to provide infrastructure, it also imposes an obligation to provide it to the extent that the requesting party can use it competitively.

So, what impact does this regulation have on competition law? Essentially, through this regulation, the CBRT, as the regulator of the relevant market, establishes that all PSP infrastructures are essential facilities for other PSPs. When drafting the legislation, the CBRT did not set any criteria for which PSPs are obliged to share infrastructure. In other words, although the CBRT could distinguish between PSPs with a market share or payment volume above a certain level, it has imposed this obligation on all PSPs regardless of these criteria. This regulation’s impact on competition law is that all infrastructures are defined as essential facilities for PSPs and, accordingly, as separate markets. Once this market definition has been established, it is necessary to assess whether the relevant bank’s actions fall under the category of refusal to contract, margin squeeze, or a different category of infringement. Therefore, the CBRT, the sector regulator, defines the border of the market through ex-ante regulation.

Conclusion

The PFs and banks compete in the merchant acquiring market (this is the market definition adopted by the TCB, see). In other words, merchants can receive POS services from either PFs or banks. For this reason, banks may be motivated to prevent the growth of PFs. There are two main reasons for this. Firstly, as the PFs enter into agreements with merchants who are existing or potential customers of the banks, the banks lose customers. This results in a loss of income for the banks. Secondly, the existence of PFs exerts competitive pressure on banks, who may be forced to reduce their prices (commission rates) for certain merchants. It is of course, not expected that banks would be content with the competitive pressure exerted by PFs. All these issues arise from competition in the merchant acquiring market.

Law No. 6493 and the CBRT’s Payment Services Regulation are both based on the fact and presumption that each credit card network (or brand partnership) is an indispensable commercial partner for PFs and merchants. Thus, one may think that each credit card network could be considered a separate market as PFs cannot compete (due to the portfolio effect) in the market without these products, especially card families such as Bonus and World. Without access to the POS of banks, particularly those with a stronger position, such as Bonus or World, they may quickly exit the market. Nevertheless, TCB’s current case-law suggest a wider market definition (i.e. the market for services related to instalment payments with credit cards and the market for services related to single payment with credit cards).

Furthermore, TCB has certain precedents where TCB consider the relationship with banks and PF as vertical thus consider the Vertical Block Exemption Communique numbered 2002/2 (“Turkish VBER”) as applicable. To put it another way, whereas the CBRT’s secondary legislation classifies the infrastructure of all PSPs as an essential facility, the Turkish VBER, which was issued by the TCB in 2002, permits providers (i.e. banks) that do not exceed a specified market share threshold (none of the banks can exceed the 30% threshold under the TCB’s current market definition) to impose non-competition obligations on buyers (i.e. PFs), as well as tying products and allocating exclusive customers (basically any restriction allowed as per Turkish VBER). It is evident that these two pieces of legislation are in direct opposition to one another, thereby engendering a substantial diminution of legal certainty. Consequently, it is reasonable to predict that the TCB may modify its jurisprudence on the definition of the market and the applicability of the Turkish VBER to the relationship between banks and PFs in the near future.

Lastly, in accordance with the prevailing market definition stipulated by the TCB, it is evident that no bank currently holds a dominant position within the industry. This complicates the application of competition law in intervening banks’ unilateral pricing behaviour. This indicates that even if banks consent to collaborate with PFs by providing necessary infrastructure, banks will retain the discretion to impose higher commission fees on PFs compared to merchants. This course of action, which is regarded as a standard margin squeeze within the context of competition law, cannot be addressed due to the absence of a dominant position. Therefore, it may be concluded that banks will be at liberty to increase the costs of their competitors in the merchant acquiring market. Yet, this will surely contradict with the Law No. 6493.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

When does the FSR filing kick-in in public tenders in the EU?

Since the entry in force of the FSR, the bidders for public contracts across EU Member States and, most recently, the EU institutions themselves,[1] are subject to mandatory FSR filing to the European Commission (“EC”) (DG GROW), in case:

• Contract value of the public contract is equal or above €250 million; and in cases where the tender is divided into lots, the aggregate value of the lots applied for by the bidder is equal or above €125 million;[2] and
• Financial exchanges with at least one non-EU government of the bidding entity (including its subsidiaries and parent companies), together with its main suppliers or subcontractors, exceed €4 million three years prior to the FSR filing (so-called foreign financial contributions or “FFCs”).[3]

The EC also has vast “call-in” powers under the FSR for public tenders below the thresholds, but, so far, it has not exercised such powers.

To date, the EC has received an unexpectedly high volume of FSR filings in EU public tenders: over 2112 in total, comprising 1734 FSR declarations and 322 FSR notifications. The FSR filings concern more than 400 public tenders in the EU, mainly from France, Germany, and Italy. To tackle the huge number of filings, DG GROW has recently formed a new “Foreign Subsidies” Unit within its Directorate G, in charge of reviewing these filings.

Who is the notifying party in the FSR filing in public tenders in the EU?

The notifying party in FSR filings is the bidding entity, including its direct subsidiaries and direct or indirect parent companies, but excluding its sister companies, unless they act as main suppliers (or subcontractors) to the bidding entity. This is known as the linear ownership structure, which applies only to bidders in FSR filings for public tenders, but not to FSR filings under the FSR M&A tool (where the notifying party is the entire group to which the buyer belongs).

The notifying party also includes the main suppliers or subcontractors of the bidding entity (even if these entities are unrelated to the bidding entity), which are also bound to submit an FSR filing and disclose information, in particular their FFCs.

How to make an FSR filing in public tenders in the EU?

To make an FSR filing, the bidding entity needs to fill in the Form FS-PP. The online form of the Form FS-PP is available here. The bidding entity should indicate whether the FSR filing takes the form of a notification or a declaration  in the form, as it determines which sections of the Form FS-PP need to be completed. Certain sections of the Form FS-PP – such as FFCs and supporting documents – may be provided as annexes.

The FSR filing is made to the contracting authority, which calls for the tender bids. However, the only authority which can review the FSR filing is DG GROW within the EC. The contracting authority is thus solely a postbox for the FSR filing, although an important one, as FSR filings cannot be formally submitted directly to the DG GROW.

In the FSR filing, the bidding entity provides information on the public tender itself, its linear ownership structure, including whether it is participating in the public tender together with the others and whether it relies on the capacities of other entities to meet the selection criteria (i.e., main suppliers or contractors). Most importantly, the bidding entity provides information on the relevant FFCs as well as the financial statements of the past three years. While the narrower scope of the linear ownership structure facilitates the identification of the FFCs, it remains the case that the FFC data collection takes time and internal coordination within the companies, as well as external counsel consultation.

The main suppliers and contractors of the bidding entity are also bound to disclose their FFCs. In case the FSR filing includes multiple unrelated parties, this raises a number of challenges concerning the way in which this confidential information is gathered and shared to the authority. To address confidentiality concerns and prevent sensitive information from being disclosed to the contracting authority, consortium partners, and/or main suppliers (or subcontractors), companies may opt for password-protected submissions.

FSR review under the public tender tool

The FSR review by DG GROW starts once the contracting authority transfers the FSR filing to DG GROW. Although expected to do so “without delay”, in practice, the transfer is often left at the contracting authority’s discretion.

The EC’s review of the FSR filings could be done in two phases:

• Preliminary FSR review phase (akin to Phase I in merger proceedings). In single stage tender procedures, the Phase I FSR review lasts for 20 working days (WD), extendable only once by 10 WD. In multi-stage procedures, the FSR may need to be filed twice for the same tender.
• In-depth FSR review (akin to Phase II investigations in merger proceedings). If the EC finds “sufficient indications of subsidies” that could distort the EU internal market, it will adopt a decision to open an in-depth investigation. In open procedures, the Phase II review lasts for up to 110 WD from complete notification, extendable only once by 20 WD in exceptional cases. In multi-stage procedures, the Phase II review lasts for 90 WD from the updated notification, extendable only once by 20 WD in exceptional cases.

To date, the EC has opened three in-depth investigations,^[4] which were not pursued until the end, as the bidders withdrew their offers, as a response to the investigation. The EC has, most recently, also declared two FSR filings incomplete, which dismissed the bidders from the tender process.

Key Insights

• The FSR filings for public tenders differ significantly from those under the FSR M&A tool. The FSR filings in EU public procurement are reviewed by DG GROW; the disclosures apply to the linear ownership structure of the bidding entity; include disclosures from the main supplies or contractors of the bidding entity; and the FFCs thresholds are lower in FSR filings under the public tender tool.
• The contracting authorities often provide detailed instructions on how the FSR filing should be done as part of the tender offer (and these may also differ between contracting authorities). You should check carefully the tender documentation concerning any FSR filing requirements. Note that the FSR filing obligation applies even if it is not mentioned in the tender documents. If the relevant FSR thresholds are met, you should carry out a self- assessment to rule out whether a FSR filing is not needed.
• The preparations for FSR filings are burdensome and time-consuming. Make sure you have all the information needed for the FSR filing, and that your chosen consortium members, main suppliers (and subcontractors), do too, as they will be also subject to FSR filing disclosures. Do not count on waivers to achieve completeness, as so far, DG GROW has been reluctant to grant any waivers. DG GROW is carefully scrutinising FSR filings and the parties should consider consulting specialised FSR counsel to avoid any delays or complications in the FSR review, which risks derailing the bid review and selection process.
• Make sure to ensure confidentiality of your FSR information vis-à-vis your partners in the tender, but also towards the contracting authority.

Anastasios Tsochatzidis (White & Case, Trainee, Brussels) contributed to the development of this publication.

[1]             Pursuant to EU Financial Regulation No. 2024/2509, which applies as of 30 September 2024.

[2]             There is a narrow group of public contracts, which are exempt from FSR filings. These categories mirror the exemptions under which Member States are allowed not to pursue EU public procurement rules.

[3]             Companies below the FFC threshold of €4 million are still under an obligation to attest it by submitting an FSR declaration to the EC. The notion of FFC is very broad and includes a variety of transfers of funds or liabilities, such as capital & equity contributions, grants, State loans & guarantees as well as tax exemptions / incentives. The FSR also allows for exempted categories of FFCs, as a result of which, the disclosures made in FSR filings should be examined, if possible, with external FSR counsel.

[4]             Cases FSP.100147, FSP. 100151 and FSP. 100154.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

Our previous blog posts on the revision of the FDI Screening Regulation can be assessed here, here, here, or here.

To recap

The 2019 FDI Screening Regulation was the first legal instrument on the EU level for FDI screening. It aimed to:

• address the (geo)political and economic impact of shifts in global power distribution,
• respond to international regulatory competition in FDI screening (e.g., from the US, Canada, Australia, and Japan); and
• tackle problems of decentralisation, fragmentation, and diversity in FDI screening at the Member State level.

Based on Art. 207 TFEU, its main innovation was the establishment of a cooperation framework for FDI screening between the European Commission and the Member States. This cooperation mechanism allows for the exchange of information among Member States. It further enables the European Commission to issue non-binding opinions when an investment threatens the security or public order of more than one Member State or when it could undermine a strategic project or programme of interest to the entire EU. Under the 2019 Regulation, FDI screening remained the exclusive responsibility of the Member States. While the creation of a screening mechanism was voluntary, Member States were required to provide details about the features of existing or new mechanisms. Additionally, they were free to determine the scope, coverage, and procedural requirements of such mechanisms.

A 2022 OECD Report, a 2023 Special Report of the European Court of Auditors, and several annual reports as well as extensive evaluation of the European Commission named shortcomings of the existing 2019 FDI Regulation. Criticism has centred on the lack of FDI screening mechanisms in some Member States. Greece, for example, only joined recently in May 2025, leaving just Cyprus and Croatia without an FDI screening system (with Bulgaria lacking an implementing system and thus being de facto without FDI screening). The scope, procedures, and timelines varied on Member State level – and the Xella problem remained (see our blog here). This situation leaves the EU resembling a patchwork quilt – recall “A Chain Works Just as Well as its Weakest Link”.

The European Commission proposal

Consequently, the European Commission proposed a revised Regulation on 24 January 2024, along with five initiatives “to strengthen the EU’s economic security at a time of growing geopolitical tensions and profound technological shifts.“ In an effort to further harmonise FDI screening, the new Regulation is intended to be based on both the common commercial policy under Article 207 TFEU and the internal market under Article 114 TFEU.

To summarise the draft, the European Commission proposes:

• Mandatory screening for all EU Member States,
• Minimum sectoral scope listed in Annex I and II, consisting of 1) Projects or programmes of Union interest (e.g. the Union Space Programme, Horizon 2020, IPCEIs, etc.), 2) a common list of dual-use items subject to export controls, 3) the common military list of the European Union, 4) critical technologies, such as advanced semiconductors, artificial intelligence, and quantum technologies (which, according to the Commission, should in the final regulation be more or less aligned with the sectors covered by its Outbound Investment Recommendation of January 2025), 5) Critical medicines, and, 6) critical entities and activities in the Union’s financial system (eg central counterparties, payment systems and institutions, central securities depositories, etc.)
• Coverage of indirect intra-Union investments, i.e. intra-Union investments by investors ultimately controlled by third-country persons (addressing the Xella problem),
• Minimum harmonisation of Member States procedures, such as timelines, standstill obligations, two-phase investigations, judicial review of decisions, etc.,
• Enhanced information exchange and accountability via the cooperation mechanisms, including a duty to state reasons when not following a Commission opinion
• An own initiative procedure for Member States (for investments in the territory of other Member States) and even the European Commission, where foreign investments were not notified under the cooperation mechanism, if they believe the investment may harm security or public order in that Member State or the EU.

The European Parliament’s position

Largely relying on the Committee on International Trade’s report and amendments, the European Parliament adopted its position for the upcoming trilogue on 8 May 2025. The Parliament sees the urgency against the current geopolitical climate and the definite need to stand closely together on European Union level.

Consequently, but overall surprisingly, the Parliament wants to go even beyond the Commission’s proposal by:

• Widening the sectoral scope to include media services, critical raw materials, the transport industry and infrastructure, electoral infrastructure, and very large farms (>10.000 hectares), as well as expanding the definition of some sectors mentioned in the Commission draft, such as artificial intelligence or semiconductors,
• Increasing the harmonisation of national procedures, including mandatory screening of certain greenfield investments,
• Granting the Commission investigation and decision-making powers on its own initiative in cases where an investment is suspected of affecting the security or public order of more than one Member State or in cases of disagreement between Member States regarding potential security or public order risks,
• Introducing a single EU e-portal for all filings at national level.

The Council’s position

The Council adopted their negotiating mandate last week, on 11 June 2025. Naturally, an instrument dealing with economic and national security is quite close to the hearts of the Member States. Due to the differing systems and experiences at the Member State level, the Council pulled back to some extent from the European Commission’s draft, maintaining its spirit, but clearly falling short of the far-reaching position taken by the European Parliament. The Council largely focuses on establishing a minimum necessary standard while encouraging Member States to go further within their national systems.

In their mandate, the Council includes:

• Continued mandatory screening, including the screening of indirect intra-Union investments,
• A limitation of the scope to 1) projects or programmes of Union interest (with the removal of certain items, such as Horizon programmes), and 2) critical technologies, such as advanced semiconductors, artificial intelligence, and quantum technologies (with some changes to the definitions),
• An emphasis on the exclusive and final decision-making power of the respective Member States, cancelling any own-initiative procedure by other Member States or the Commission,
• A proposal for a database maintained by the Commission, containing the outcomes of assessments under national screening mechanisms.

Comments

After the kick-off meeting of the Trilogues today, where common ground will be established, the three institutions now face tough negotiations in the coming months. Although the upcoming Danish Council presidency aims to conclude the negotiations by the end of its mandate later this year, the significant differences, especially between the co-legislators, will make finding a compromise a challenging endeavour.

All three legislative institutions recognise the underlying principle of open strategic autonomy that has guided the FDI Screening Regulation from the outset: while Europe needs foreign investment and seeks to remain a reliable destination for it, protecting vital security interests, both of the Member States and of the EU as a whole, is essential. The urgency is clear, and the lessons of history are well understood, given the recent loss of reliable partners and the wars both within Europe and on its doorstep. Still, the investors whom the EU continues to consider reliable need predictability – and they need it sooner rather than later. Even with an agreement by year’s end, a new FDI Screening Regulation is unlikely to be fully operational before 2027, more likely 2028.

Looking at the different mandates of the three institutions, we encounter the classic subsidiarity challenge in the European Union: what is best left to the Member State level and where does harmonised EU action adds value? In the context of FDI screening, this raises a further issue: Member States have – and should retain – the responsibility to safeguard their national security under Article 4(2) TEU and to protect their essential security interests under Article 346 TFEU, which is closely tied to their national sovereignty.

At the same time, Member States may have different interpretations of what constitutes a security risk, creating potential vulnerabilities for other Member States and for the Union as a whole. And here we arrive at the crucial point: what are the common security interests of the Union as a whole? How many Member States must be affected, and what types of security risks are included to mandate action on Union level? The European Union still lacks a clear definition and shared understanding of the concept of common EU security interests. Without it, a truly multi-level European screening mechanism (similar to EU competition and merger control) – where national security interests are screened at the Member State level and common EU security interests at the EU level – remains a distant goal. For international investors, however, the internal market is the EU’s key asset and an attractive investment opportunity. Protecting the internal market as a viable investment ground would be the logical response and should be at the heart of the upcoming negotiations.

**

This blogpost includes insights gathered at the 2025 BDI/CELIS German Chapter Investment Screening Conference, the 20th Annual Conference of the GCLC on “Present and future role of competition policy and the EU’s open strategic autonomy“, the 2nd Conference on Economic Security and Market Regulation,  and the 3rd Annual Conference on Investment Control in Austria and the EU. However, all thoughts shared here are my own and cannot be contributed to anyone else.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

On 15 May 2025, AG Emiliou delivered his Opinion on the request for a preliminary ruling submitted by the Portuguese Tribunal da Concorrência, Regulação e Supervisão in Tondela et al. 

The Opinion was eagerly awaited, as it represents the first occasion on which a no-poach agreement – an understanding between employers not to solicit or hire one another’s employees – has undergone in-depth scrutiny before the EU courts. The European Commission had already left its mark on the subject, first through its 2024 Competition Policy Brief on antitrust in labour markets and, subsequently, through its decision of 2 June 2025 imposing fines totalling EUR 329 million on Delivery Hero and Glovo, inter alia for their participation in a no-poach agreement. 

AG Emiliou’s Opinion is therefore welcome, as it confirms a number of assumptions concerning the application of EU competition law to labour markets. Yet, like Giorgio Monti, I find parts of the reasoning in Tondela somewhat disconcerting. Several of the AG’s conclusions merit a second look, and the interested reader might appreciate a touch more nuance when EU competition law is transposed to labour-market contexts. 

Portuguese Football in April 2020 – When the Clock ran past 90 

To explain the puzzlement, I shall first set out the key facts. 

As in many European leagues at the start of 2020, Portugal’s first and second divisions suspended all fixtures when the COVID-19 pandemic struck. It quickly became clear that the season might have to extend beyond its scheduled end date of 30 June 2020 so as to accommodate the full programme of 2019/20 matches. The clubs realised that this extension could clash with roughly one-third of the players’ contracts, which were due to expire on 30 June 2020, leaving them at risk of fielding depleted squads once those contracts lapsed. 

They also foresaw grave cash-flow problems. If their worsening finances prevented them from paying future wages, players might refuse to play or even terminate their contracts for just cause – whether owing to non-payment or to non-deployment following the cancellation of matches. In turn, clubs with better financial funding could soak up players from clubs suffering from the financial ramifications of the pandemic. 

In an effort to resolve the impasse, the Portuguese national football association opened talks – principally about extending players’ contracts – with the Portuguese union of professional football players. No accord was reached (see the summary of the request for a preliminary ruling in Case C-133/24, paras 16 et seq.). 

Subsequently, 31 Portuguese clubs – every first-division club and most from the second division – together with the Portuguese national football association, took matters into their own hands. On 7 and 8 April 2020 they agreed not to sign any player who unilaterally terminated his contract owing to COVID-19-related issues or to an exceptional measure adopted in response to the pandemic, in particular a decision to extend the 2019/20 season. 

The object of the agreement was to hold players to their contracts until the conclusion of the prolonged season and to deter terminations for just cause. It was also intended to preserve the competitive integrity of the Portuguese leagues and to stave off sector-wide financial harm. In essence, the agreement aimed to shield clubs from the twin threats of mass contract terminations and lay-offs triggered by an inability to meet salary obligations (the summary of the request for a preliminary ruling in Case C-133/24, paras 38, 40). 

Object or Effect? Picking Sides for No-Poach Agreements 

The most notable finding of AG Emiliou is that no-poach agreements such as the one concluded between the Portuguese clubs may amount to restrictions only “by effect”. I shall return to that point below. Like the AG, I prefer to begin with the notion of restrictions of competition “by object.” 

AG Emiliou takes the view that no-poach agreements fall prima facie within the by object category (Opinion, para 49). Before reaching that conclusion, he embarks on a brief detour through the elusive concept of “restriction by object”, tracing its evolution in the case-law and charting how it has settled into the form we know today (Opinion, paras 26 et seq.). He then observes that no-poach agreements “have all the characteristics to be considered prima facie restrictive of competition” and therefore generally belong in the “by object” category (Opinion, para 49). In particular, they would appear to fall under Article 101(1)(c) TFEU, as they entail the sharing of a supply between competitors – here, the supply of labour (Opinion, para 50). 

This is the first point on which a reader might crave a touch more nuance. The AG notes that classifying no-poach agreements as restrictions by object applies to (1) “agreements in which two or more undertakings agree not to hire or solicit staff from each other”, (2) “at least when entered into between actual and potential competitors” (Opinion, para 49). Both limbs warrant closer scrutiny. 

Floodlights on Labour Markets – and their Intricacies 

I shall begin with the second limb, for the notion of (potential) competitors on labour markets – and thus market definition – is of wider significance. I concur with AG Emiliou that the relevant market in Tondela is, in all likelihood, the market for recruiting professional footballers (Opinion, para 64). Yet, while the AG confines himself to a terse delineation, defining markets for the recruitment of labour presents several challenges that warrant a step back. 

Because competition law is largely designed around co-ordination on the supply side, defining markets where employers compete for employees – i.e. the upstream demand side – requires a change of perspective. Instead of examining demand-side substitutability, one must focus on supply-side substitutability: put simply, which jobs would employees regard as substitutes? 

This shift forces us to confront the elusive nature of workers’ preferences. Labour is a peculiar “good,” inseparably tied to human suppliers whose priorities often (or even predominantly?) extend well beyond raw economics – think elements such as work-life balance or reputation. Literature recommends the SSNDW test (small but significant non-transitory decrease in wages) to delineate labour markets, yet a wage-centric lens may fail to capture the full range of supply-side considerations. 

In any event, geographic scope poses an even stiffer challenge. Studies suggest that employees’ willingness to relocate for work is surprisingly low, though mobility varies across workforce segments: labour markets for low-skilled workers tend to be local or regional, whereas those for highly skilled employees may span far wider areas. 

Absent further research, one would assume that the labour market for professional footballers stretches beyond national borders. Certainly, the regular stream of high-profile transfers between Europe’s top clubs suggests considerable mobility – even if only at the margins of top players. 

Non-Hire vs Non-Solicit: Different Formations 

Turning to the first limb, what struck me most was AG Emiliou’s statement that every “no-poach agreement” possesses all the hallmarks of a prima facie restriction of competition. 

A finer distinction is required. As he notes, no-poach agreements come in two broad guises: non-hire and non-solicit agreements. The former undoubtedly resemble supply sharing within the meaning of Article 101(1)(c) TFEU and display the traits of a restriction by object. The latter, however, present a far less straightforward picture. 

Non-hire agreements may indeed be considered inherently harmful to the competition for talent. At least where all – or nearly all – employers on a given labour market adopt such an agreement, employee mobility is driven to zero and competition grinds to a halt. Workers are locked into existing contracts and deprived of the leverage to secure better terms as they are unable to make credible threats to change employers. Downstream, this sub-optimal allocation of human capital produces detrimental knock-on effects on efficiency and innovation (see Opinion, para 52). 

By contrast, the ramifications of non-solicit agreements are more nebulous. Because employees remain free to approach competitors of their current employer, such clauses do not invariably “freeze” the labour market. Workers can, in principle, still threaten to move, preserving much of their bargaining power. Accordingly, non-solicit provisions do not necessarily present the inherent competition risks required to label them restrictions by object. Their impact is highly context-dependent. 

Two considerations, in particular, merit attention when assessing a non-solicit agreement: 

• First, in labour markets with high transparency – where demand is advertised openly (for instance, via online job boards) – the chilling effect of non-solicit clauses may be modest. Conversely, in markets where vacancies are rarely publicised, the effects of non-solicit agreements may be more akin to those of non-hire agreements. If employee mobility is already dampened because posts are not openly advertised, a non-solicit clause may be the proverbial straw that breaks the camel’s back. 

• But second, even then, the impact of non-solicit agreements can remain limited. Highly skilled workers, who generally grasp their own “market value” and know their potential alternative employers, are less likely to be deterred from approaching a rival employer than employees with scant knowledge of market conditions. 

These observations are hardly exhaustive, but they should suffice to show that treating every species of “no-poach agreement” as prima facie restrictive by object is an oversimplification. 

Throw-In to Holism: The AG’s Detour towards the “by Effect” Part of the Pitch 

I must now address the elephant in the room: AG Emiliou’s eventual finding that the no-hire agreement restricted competition by effect, not by object. 

Although the Portuguese clubs plainly sought to halt competition for footballers, the AG focused on the agreement’s legal and economic context and on its stated aims. He concluded that it should not be deemed a restriction by object, because its genuine purpose was to safeguard the fairness and integrity of the sporting contest disrupted by the pandemic (Opinion, para 71). That reasoning is noteworthy. 

As Giorgio Monti rightly pointed out, conflating the impact on the upstream labour market with broader welfare considerations in the downstream market for professional football sits uneasily with the architecture of Article 101 TFEU – particularly the calibrated interplay between paragraphs (1) and (3). There is, I submit, no basis for treating agreements differently merely because they relate to labour markets. 

One wonders whether the AG’s “holistic” approach rests on the notion that no-hire agreements warrant a bespoke competition analysis simply because they touch matters related to labour markets. In para 69 he remarks that he would not be “against taking into consideration” “issues relating to labour […] within a competition analysis”, yet in the same breath concedes that the interplay between “labour issues” and “competition issues” remains unclear, especially where an Article 101 TFEU case concerns “alleged harm caused to workers”. This suggests a view that agreements affecting labour markets are a special breed, demanding a novel application of competition law. 

Those remarks did, in fact, not just puzzle me. In my view, they convey a line of reasoning that points in the wrong direction. Article 101 TFEU cases are not brought for “alleged harm to workers”; they are brought for anti-competitive conduct in labour markets, which, in turn, harms workers. In the same vein, and on a more general note, there are no “labour issues” in anti-competitive practices affecting labour markets – only competition issues. The intricacies of labour markets (and of labour law) must not obscure the basic truth that these markets, like any other, thrive on competition – which is why we in the antitrust community are talking about them in the first place. 

Conclusion before the Full-Time Whistle 

Tondela underscores that policing employer conduct in labour markets – no-poach agreements foremost among them – demands sensitivity to those markets’ specific dynamics. Yet, I remain convinced that the existing EU competition law framework is more than equal to the job; the real task lies in mapping established principles onto labour-market nuances. While this can be quite the intellectual challenge, it is one that we must face in order to maintain and foster competition in line with the cardinal principles of EU competition law – including in labour markets. And I do hope that the ECJ’s ruling will underscore this sentiment. 

* I have never acted for any of the parties involved and am in no way affiliated with them. The views expressed here are my own and do not necessarily reflect those of Hogan Lovells International, its clients, or its personnel. 

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

Particular attention should be given to this precedent in the fast-moving start-up scene, with IoT investments and in financial services, where comparatively large investments in (at least initially) non-controlling stakes are often secured through board positions and the like.  The message of Delivery Hero is clear: as long as there is no sole control achieved, companies remain independent.  Care needs to be taken to avoid a premature management creep in aligning labour law approaches and commercial positioning.

Dual focus: minority stakes, no poach

The Delivery Hero decision is of key importance for two reasons, as it formalizes the EU’s position regarding the facilitation of a cartel through:

1. The functioning of labour markets with no poach arrangements (i.e. arrangements not to hire or actively approach each other’s employees); and
2. The use of minority stakes to remove competitive constraints.

Both aspects of Delivery Hero will require amending or at least fine-tuning compliance when it comes to exchanging sensitive information during and after acquiring a minority stake (read: clean team rules, who should sit on the board, what can be discussed at the board / any special step-out or reporting provisions, etc.).

Four (expensive) years to becoming one company: Delivery Hero’s slow path to control

Delivery Hero and Glovo are both active in the online food delivery sector, Delivery Hero is by origin German, Glovo Spanish.  In July 2018, Delivery Hero took a 15% non-controlling stake in Glovo, which was progressively increased through subsequent investments, setting the two companies on course to full merger in 2022.  The latter was cleared by the Spanish Comisión Nacional de los Mercados y la Competencia (CNMC) in February 2022, see here, whereby Delivery Hero acquired full control over Glovo, near four years after the initial stake.

The Commission found that, by the time Delivery Hero acquired sole control over Glovo (in July 2022, upon consummation), the two companies had already and pre-emptively so removed progressively the competitive constraints between them and “replaced competition with multi-layered anticompetitive coordination” (as per the EU press release as referenced above).  The Commission found that during this period, the two companies had agreed:

• Not to poach each other’s employees;
• To exchange commercially sensitive information; and
• To allocate geographic markets.

Following two Commission dawn raids (in June 2022 and in November 2023) at Delivery Hero and Glovo premises, two years of investigation, and the issuance of a Statement of Objections in July 2024, the companies are set back EUR 223 million (Delivery Hero) and EUR 106 million (Glovo) through settlement.  While the EU press release provides scant detail on the fine calculation, these absolute numbers show that the Commission attached a material gravity to the infringement, only applying the 10% settlement discount.

Side-note, the Commission takes great care to identify the Delivery Hero investigation as an “own-initiative inquiry”, prompted by information received “from an (unnamed) national competition agency” and “via the anonymous whistleblower tool”.

Let’s look at the dual focal points in some more detail in the following.

No poach

No poach as an antitrust concern

No poach as an antitrust concern emerged in the U.S. in late 2016, when the U.S. Department of Justice and the U.S. Federal Trade Commission issued their joint “Antitrust Guidance for Human Resource Professionals” (as updated in January 2025).  The first criminal indictment followed in January 2021, against Surgical Care Affiliates (“SCA”).  In Europe, while some national competition agencies where quicker to pick up no poach, it took till October 2021 for then-Competition Commissioner Vestager to pronounce the Commission’s big picture view on no poach, in her “A new era of cartel enforcement” speech.  In that speech, she states:

• “Some (…) cartels do have a very direct effect on individuals, as well as on competition, when companies collude to fix the wages they pay; or when they use so‑called ‘no‑poach agreements’ as an indirect way to keep wages down, restricting talent from moving where it serves the economy best”;
• “And that’s not the only way that an agreement not to poach each other’s staff can create a cartel. There are markets where you can only compete if you have expensive machinery, or costly IP.  And then there are those where the key to success is finding staff who have the right skills.  So in these cases, a promise not to hire certain people can effectively be a promise not to innovate, or not to enter a new market”.

Ms. Vestager’s speech in late 2021 was eagerly anticipated, finally positioning the EU after a long silence in Europe vis-à-vis no poach (while the U.S. was busy making antitrust headlines).  Accordingly, her statements were widely reported at the time – but seemingly not within the online food delivery sector, where by then Delivery Hero and Glovo had not only concluded a shareholders agreement (formalizing the initial investment in July 2018) containing a limited reciprocal no-hire clause for certain employees, but shortly after expanded this agreement to a general arrangement not to actively approach each other’s employees.

The Commission’s read of no poach in Delivery Hero

In the press release now accompanying the decision, the Commission identifies as the central issue with no poach “fewer job opportunities for employees” and positions its investigation into Delivery Hero to contribute to “ensuring a fair labour market where employers do not collude to limit the number and quality of opportunities for workers but compete for talent”.  Competition Commissioner Ribera, during her remarks presenting the decision, added that no poach agreements are a “purchasing cartel” where companies no longer compete for labour input (and thus harm citizens directly by suppressing wages and reducing labour mobility; “competition rules matter to citizens’ daily life”).  She also held that no poach arrangements impede an industry’s overall performance by lowering productivity and stifling innovation.

Relevance for compliance training

Delivery Hero will require special training modules for HR professionals what to say, or not to say, how to formulate labour agreements, etc.  Co-opting no poach arrangements into the shareholders agreement – Delivery Hero-style – is certainly not advisable.

In this regard, compliance professionals will be well-advised to be mindful of the Commission’s admission that in reaching its decision in the Delivery Hero case, it was assisted in its investigative efforts by a whistleblower using the anonymous whistleblower tool available on the EU website.  Disgruntled (former) employees may well spill the beans on practices which they found limitative, triggering regulatory interest.

In the international context, calibrating compliance efforts between major jurisdictions will remain complex, certainly where dispatch employees congregate in particular areas (think financial hubs, silicon hubs).  Different “sending” jurisdictions will cover their respective labour contracts, and those need to “fit” the requirements of the “receiving” jurisdiction – now increasingly also with an antitrust component.  While the approach to no poach has not yet settled (the U.S. might have crested after the original indictment in SCA was dismissed in November 2023; the EU may or may not end up treating no poach as a true “by object” infringement devoid of any effects assessment, as is also at stake in Tondella), expert advice would seem advisable, and regular updating, too.

Minority stakeholder conduit for cartelization

Minority stakes as facilitators

Best to start off on the second focal point of Delivery Hero with Commissioner Ribera’s remarks, where she held that “of course, owing a stake in a competitor is not illegal in itself”.  The problem emerges (only) when that stake is “used to gain insight information and influence decisions in ways that can harm competition”.

The simple read is, short of two competitors becoming one company, and instead remaining independent market players, the EU’s far-reaching by object interpretation of exchanges (viz., no need to agree) of sensitive forward-looking information being on par with actual fixes of prices, quota, capacities, costs, territories, customers and the like remains in full force.

The harder read is, how did Delivery Hero and Glovo cross the thin red line?  Commissioner Ribera states that they exchanged sensitive information “beyond what was needed for a corporate investor to protect [or, at a different place in her remarks, to monitor] a financial investment decision”.  What do we learn about the “beyond”, where did a premature management creep go too far:

• Delivery Hero and Glovo apparently shared “strategy documents” and held “meetings to share knowledge” including on “current or future commercial strategies, (and) new offers”;
• While the EU treats the market allocation aspects as a third strand of the infringement in its decision, systematically, these aspects also emerged due to the ability the minority stake gave Delivery Hero to “convince Glovo to share markets in (..) two ways: (i) directly, by using or threatening to use its approval rights over specific decisions, and (ii) indirectly, by influencing other Glovo shareholders”. Effectively, Delivery Hero and Glovo discussed and agreed which of them would enter new national markets where neither had a presence yet, and they also refrained from entering markets where the other was already present.

Relevance for compliance training

Minority stakes will (and should) continue as a means for investment, as they:

• Allow companies to collaborate strategically without triggering a full acquisition or merger;
• May support joint R&D, co-development of products, or technology sharing, especially in the fast-moving start-up scene, in industries like IoT, biotech, or automotive, and in financial services;
• And do so in a commercially less risky and more flexible manner than a merger.

While not leading to control, minority stakes are often secured through board positions and the like, and – as in the case of Delivery Hero and Glovo – they also allow a more cautious managerial decision-taking as to whether or not a more fulsome integration makes commercial sense over time.

Where, then, is the antitrust limit of “co-opetition” (compare the book, same title, by Brandenburger and Nalebuff, 1996), or, how best to advise deal makers to put their minority stake in place without risking the EUR 329 million fine Delivery Hero and Glovo accrued on top of their investment?  Some key compliance pointers based on experience dealing with the general risks attached to sensitive information exchanges between competitors in the EU include:

• Use strict clean team rules when negotiating;
• Install clear and upfront protocols (ideally in the shareholders agreement) for the functioning of the board, the parameters of information to be discussed / not discussed within the remit of the minority stake, a “step out” process to avoid spoliation, etc.; and
• Appoint individuals to board positions that are less likely to be at risk, and provide them with guardrails alike targeted compliance training and structural firewalls to lower exposure.

Afterthought: beware the social media treasure trove

For good measure, Delivery Hero also serves as a cogent reminder that social media / messaging services (alike WhatsApp, Telegram, Facebook Messenger) are no safe haven from antitrust enforcement.  If anything, Delivery Hero shows that ever since the Nexans judgment, EU regulators have become quite keen on getting their hands on private social media or personal messaging accounts of employees, when (also) used for business purposes.  As useful all-in-one compliance training illustrations go, also plenty of the evidence regarding market allocation and market entry in Delivery Hero initially stems from WhatsApp chats.

*All views are the author’s and do not represent those of clients.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

Advocate General Emiliou delivered three interlocking Opinions concerning decisions of football governing bodies. Two of these cases were brought to challenge elements of the FIFA Football Agent Regulations (Case C-209/23, RCC Sports and Case C-428/23, Rogon). The third concerns a decision of the Portuguese Competition Authority which had found that a no-poach agreement by football clubs during the Covid pandemic restricted competition (Case C-133/24, Tondela). The Opinions are relevant not only for the sports sector but more generally for an understanding of basic principles of EU competition law. This blogpost focuses on the manner in which the Opinions frame the justifications available to undertakings and associations of undertakings whose conduct infringes competition law and internal market law. 

Some fundamental premises should be stated at the outset (RCC Sports, paras 29-33). It is lawful, as a matter of EU Law, for a private association (like FIFA or a national football association) to set up rules governing its activities and those of economic actors affiliated to it. However, if these rules harm competition, then Article 101(1) TFEU applies; and if they have an adverse effect on the internal market, so do the free movement provisions (e.g. if the rules hamper the provision or receipt of services contrary to Article 56 TFEU). There is no automatic immunity for anticompetitive rules issued by sport governing bodies. However, on a case-by-case basis, associations may justify restrictive regulations. A brief remark on the way the theory of harm is articulated in these Opinions will be helpful first before discussing justifications. 

Theories of Harm 

In RCC Sports and Rogon, the rules (e.g. those limiting agent remuneration) may have an adverse effect on competition in two markets: the market for the services provided by agents (where competition among agents may be distorted) and the market for the transfer of players (where competition among football clubs may be distorted). The AG suggests that the test to determine if such rules are anticompetitive rests on asking whether they limit ‘the ability and incentive of those market operators to compete against their rivals’. When it comes to competition between agents, the advice is to consider whether the rules limit agents’ incentives to provide services or decrease their incentives to invest in improving their services. As regards competition between football clubs, the question is if the rules prevent clubs from offering higher fees to secure their services (RCC Sports, paras 65-67). 

Looking at these same rules from an abuse of dominance perspective, they serve to leverage one’s dominance in the market for the organization of football competition. The abuse could be exploitative by imposing unfair trading conditions or exclusionary insofar as they restrict market access to agents and thus limit competition in the two markets. (RCC Sports, paras 114-142) 

Note that there is no mention of consumer welfare. Under Article 101 and under a theory of exclusionary abuse, the harm is to the competitive process demonstrated by showing a reduction in allocative or dynamic efficiency. Under an exploitative abuse perspective, the harm is to the economic actors whose welfare is diminished, a distributive justice consideration. 

In contrast, the analysis of no poach agreements in Tondela is puzzling. At first, the AG suggests that conduct can restrict competition by object when the labour side of the market is harmed. No poach agreements mean that employers do not compete to offer better conditions for workers, resulting in a ‘suboptimal allocation of human resources.’ (Tondela, para 52)  Up to here the analysis is comparable to that in  the football agents cases: does the agreement harm the other side of the market by reducing efficiency? 

However, when looking at the specific facts of the case (i.e. that the no poach agreement was a temporary measure designed to make sure that for the final stretch of the football season rich clubs did not go on a shopping spree of players from poorer clubs) the AG suggests that the agreement would be seen as pro-competitive because it prevented the richest clubs from acquiring all the best players and thereby made the remainder of the football season exciting. If football transfers had been allowed, some clubs would definitely go down while the rich ones would be at the top of the table, having bought all the talent. A predictable league would harm the interests of supporters in watching matches, of sponsors in funding football clubs, and of broadcasters to pay to show matches. This might have led to reduced investments and ‘a lose-lose situation for clubs and players.’ (Tondela, para 65) Furthermore, the AG criticises the competition authority for not providing a meaningful answer to questions about the harm caused to consumers (Tondela, para 69). This leaves me a bit puzzled as to what the correct theory of harm should be in cases involving the exercise of monopsony power. Is it sufficient to show harm to the labour side of the market? (A partial equilibrium analysis.)  Or must one look at the overall impact on both workers and economic actors on the other side of the market? (A general equilibrium analysis.) Or should one trace any anticompetitive effects to final consumers? For what it’s worth, in my view it should suffice to prove harm to one side of the market and there is no need to trace this harm back to final consumers or to carry out some sort of overall welfare standard by which all effects are weighed up. We have Article 101(3) for an holistic analysis. More generally, a restriction of competition is about finding harm in a relevant market and when Article 101(3) is engaged, it is countervailing effects on that market which matter. We don’t do total welfare in EU competition law. 

Recall that challenges to the agent regulations were also brought under Article 56 TFEU on the basis that they restrict the freedom to provide services. Article 56 applies to a sports federation because they are in a position to impose on a group of individuals (in casu, football agents) rules that adversely affect their capacity to exercise their fundamental freedoms. Triggering this prohibition is much simpler than applying competition law. There is no need to define markets, find market power or show anticompetitive effects.  It suffices to show that the rules make it more difficult for service providers and recipients to, respectively, provide and obtain services outside their jurisdiction. Arguably this is justified because the economic power of a football association where membership is compulsory is such that one can presume that its rules harm economic efficiency (RCC Sports, paras 144-145). 

Justifications  

There are four pathways that may be pursued by undertakings who seek to justify anticompetitive practices. The AG discusses three of  these fully and brings some novelties to each. To foreshadow the analysis, note that since multiple EU rules may be applied to challenge the same conduct, it is important that the justifications that may be provided are consistently applied irrespective of how the case against the defendant is framed. 

Purely sporting exclusion 

The first is to claim that the rule is purely about sport. This excludes the application of EU internal market and competition law. According to the AG this rule is an application of two principles (RCC Sports, paras 17 to 28): (1) that EU internal market and competition law are concerned with economic activities and a pure sporting rule is not directly linked to an economic activity and (2) even if there is some indirect effect on economic activities, this is de minimis. These two legal foundations do not really work well as principles: the second contradicts the first. The AG is right in saying that some regulatory choices should not require a costly discussion about their compatibility with EU Law (e.g. the number of players that each team may field). However, this is a category that is difficult to define ex ante as there may be settings when we do want to test a rule’s compatibility with the internal market even if at first blush it looks like the rule is only about sport. A better approach might be to create safe harbours where it is presumed some rules do not offend EU Law. 

Albany International

The second (of which the Albany International strand of case-law is the sole exemplar) is to claim that there is more to life than competition because the EU ranks other things more highly. On the facts of Albany International, collective bargaining between workers and employers when this concerns working conditions falls outside of Article 101 and no explicit trade-off is necessary. We tolerate anticompetitive effects in the name of protecting workers’ rights to bargain collectively, irrespective of the consequences this has for economic welfare. Somewhat bizarrely, the AG seems to subsume this precedent under sporting rules (RCC Sports, footnote 20). With respect, this is not right. The Albany International judgment creates a self-standing rule of exclusion. It is a generous exclusionary rule  because all that has to be shown is that the ‘nature and purpose’ or the agreement justifies the exclusion of Article 101(1) (See Albany International para 60). However, since proportionality is a general principle of EU Law, it probably should also apply to this exclusion.   

What remains open for discussion is what other public interests can benefit from this exclusionary rule. What about, for example, Article 11 TFEU, requiring that environmental considerations be integrated in the EU’s policies? 

The Wouters defence

In Wouters v General Council of the Dutch Bar, the Court found a third way to justify restrictive agreements. AG Emiliou summarises the standard (which he refers to as the Meca Medina test, a judgment where Wouters was applied to FINA’s anti-doping rules) by reference to Superleague: (i) they are justified by the pursuit of one or more legitimate objectives in the public interest which are not per se anticompetitive; (ii) the specific means used to pursue those objectives are genuinely necessary for that purpose; and (iii) even if those means prove to have an inherent effect of, at the very least potentially, restricting or distorting competition, that inherent effect does not go beyond what is necessary, in particular by eliminating all competition.” (RCC Sports, para 49). In discussing this justification, the AG makes a number of points worthy of comment, by which the AG provides safeguards so that this justification does not become a ‘Get Out of Jail Free’ card for self-governing bodies (Rogon, para 41): 

• It applies when public authorities have either assigned the body in question the responsibility to pursue a public interest objective or they have ‘acknowledged and accepted the body’s activity of self-regulation.’ (Rogon, para 42) However, later in the Opinion, this is diluted because a public authority’s acknowledgment may be expressed ‘by simple tolerance’ (Rogon, para 63). If Member State silence suffices, then this requirement has no meaning. However, it seems a potentially useful way of delimiting the Wouters defence because it would reveal a Member State’s interest in protecting certain public goods.
  

• It only applies if the public interest is ‘worthy of protection under EU Law.’ (Tondela, para 78) Rules preserving the health of athletes or preventing abusive, fraudulent or unethical practices vis-à-vis athletes are objectives that pursue a ‘public non-economic interest’ (RCC Sports, para 70) and qualify. But it is not clear why they are worthy of protection under EU Law. Relatedly, the AG refers to the protection of the sports ecosystem so that it works based on the principles in art 165 TFEU (RCC Sports, para 71) – on this basis one can examine if the rules improve the quality of services agents provide or improve financial transparency. This suggests that looking into the EU Treaties gives us clues about what public interests EU Law protects. If so, rules designed to secure gender equality would also count for example.  But what of a justification that cannot be traced to the Treaty? How does one decide then what is and is not worthy of protection under EU Law? We need a rule to help us recognise these interests.
  

• It only applies when the association issues rules closely related to its core activity, for example FIFA may issue rules to protect young athletes but perhaps not to protect the environment. It is not clear why not. For example, I think the International Skating Union could forbid the organization of ice skating in Qatar to safeguard the environment because of the huge environmental impact of cooling a skating rink in the desert. This seems to be a legitimate public interest that someone regulating this sport may well feel are within its competence. The AG responds that it is for the public authorities to find the optimal way of achieving an equilibrium (Rogon, para 45) between public interests. However, if so, why delegate some to self-regulatory bodies? There is a tension here between authorizing self-regulation and saying that the public interest is for the state to determine. You cannot have it both ways.
  

• It applies if the pursuit of the public interest is achieved in a transparent manner. (Rogon, paras 46 and 70-73). The AG explains that the influence of third parties in the association’s decision-making is a relevant consideration. On the facts, the more FIFA affords agents opportunities for ‘genuine and meaningful participation’, the greater the input legitimacy of the rules and the more likely are the restrictions likely to be judged necessary and proportionate. These procedures echo the indications already found in Superleague.
  

• It applies to rules that affect members of the association as well as rules that an association drafts that affect non-members provided that the services provided by those non-members have a ‘direct and significant influence’ on the core activities of the association. (Rogon, para 74) This is nothing new, as in Wouters the rules set by the Bar Council affected the economic interests of accountants.
  

• It applies when the regulations are genuinely necessary to achieve the public interest. The AG suggests this point requires proof of three aspects: (a) there is an objective need; (b) the agreement genuinely reflects the objective in a consistent manner; (c) it is suitable to achieve that objective. (RCC Sports, para 72).  The AG suggests that when testing for suitability the associations have a degree of discretion – this is not about finding out if the agreement is the least restrictive measure, rather merely to consider if a less restrictive alternative might have been considered (Tondela, para 86). One remarkable point to note here is that the AG refers to the ECN’s statement about the need for businesses to cooperate during Covid. In his view this “might corroborate the view” that the defendants “could reasonably consider that some degree of cooperation … was, exceptionally, necessary in that period.”(Tondela, para 90) This does not seem to be a useful source on two fronts. First, it does not help work out the necessity requirement:  the ECN suggests that there may need to be cooperation, it does not loosen the necessity requirement. Second, it is hard to see how a soft law document can create any sort of legitimate expectation about the need to restrict competition.
  

What all of these points have in common is that they align the Wouters defence to the analysis of justifications carried out when Member States justify restrictions of free movement. As I argued many years ago, Wouters is best seen as an application of the so-called mandatory requirements case-law in internal market law to EU competition law. 

Article 101(3) TFEU 

The fourth pathway to escape Article 101 is found in Article 101(3). There is ample literature debating the scope of Article 101(3). The most through account today is Or Brook’s Non-Competition Interests in EU Antitrust Law (Cambridge, 2022). The AG Opinion makes significant points with respect to the first and second conditions of this provision.   

The first condition is that the agreement ‘contributes to improving the production or distribution of goods or to promoting technical or economic progress.’ According to the AG this includes ‘benefits of a non-economic nature, provided that they have tangible positive repercussions on the relevant market or on markets which are strictly connected to them.’ (RCC Sports, para 93).  He goes on to explain that if a rule restricting competition serves to prevent the abusive exploitation of young athletes, then this is a positive effect that counts under Article 101(3) because it improves the functioning of the ‘football ecosystem’.  This wide approach is defended in four ways (RCC Sports, paras 88-92, but reshuffled in this list): (i) it is consistent with precedent; (ii)  there is no need to quantify the positive effect, but the association has to have adequate reasons and evidence that there is a real positive effect (in passing the AG criticises the 2004 Guidelines on Article 101(3) as being too strict); (iii) the concept of ‘economic progress’ is the lynchpin on which non-economic gains may be pleaded and this term is not a synonym of economic growth; (iv) this interpretation is consistent with the defences available in the case-law on free movement rules and a comparable framework of analysis is logical, he says, to avoid paradoxical results. (Observe the similarity of the interests protected with the list in the Wouters defence above.) 

The second condition is that ‘consumers’ (or users if you look at e.g. the French, Italian and Dutch versions of the Treaty) secure a fair share of the resulting benefits. The AG says that while football supporters are clearly included, other categories of users count. Following Superleague, this includes national football associations and players as well as TV viewers. However, the AG’s conclusion may be queried: ‘the benefits for the different types of users can be considered in combination when balancing them against the restrictive effects produced.’ (RCC Sports, para 95) This is not consistent with Superleague where the Grand Chamber was clear that a restrictive agreement must have a positive effect on each of the parties who suffer harm (Superleague, para 194).  This makes it imperative that we understand how to frame the theory of harm for Article 101(3) must reflect those theories. 

There is a further point in his discussion of the second condition to discuss: it requires that users obtain a ‘fair share’ of the benefits. According to the AG this requires that the net impact on users is positive.  However, in my view, this is mistaken. When one balances positive and negative effects, the positive effects are those in the first limb of Article 101(3), not the second. The Court understood this as far back as 1966: “This improvement must in particular show appreciable objective advantages of such a character as to compensate for the disadvantages which they cause in the field of competition.” (Joined Cases 56 and 58/64 Consten and Grundig, p.348) To illustrate: an agreement that reduces environmental pollution to the value of EUR 100 million annually (measured by increased biodiversity, better health for plants, animals and people) which raises the price of a product bought by 500,000 people annually by EUR 5 when users would only be willing to pay EUR 4 for more sustainability probably gives the users a fair share, as they are only EUR 1 worse off while the overall societal gains are greater than the anticompetitive effects. 

Internal market justifications 

Remember that the association’s rules may also be pursued for infringing Article 56 TFEU (freedom to provide services). Here we have a case-law created justification (which some EU lawyers had, confusingly, called a rule of reason!) by which the association may justify the harm to the economic freedom of businesses by reference to: (a) a legitimate objective in the public interest which is other than of a purely economic nature; (b) secured by a rule which is suitable for ensuring the achievement of that objective and does not go beyond what is necessary for that purpose. (RCC Sports, para 146) 

The public interest basket seems to include the same considerations as in Wouters. With respect to the second, proportionality -based, limb one ‘should assess whether those rules genuinely reflect a concern about attaining those objectives in a consistent and systematic manner.’ (RCC Sports, para 153) 

The AG’s take is that the genuine concern element needs close scrutiny because any justification ‘may be knocked together ex post’ to justify a restriction of the free movement rules (RCC Sports, para 154). Moreover, the AG calls for a proportionality stricto sensu test, by which one must show: (a) that there is no less restrictive alternative and (b) that the restrictions strike a fair balance between the public interest and the interests of the agents. This aligns with his reading of the Wouters justification in Superleague. Indeed, if one compares the AG’s discussion in Wouters with the analysis of justifications in the free movement case-law, there is much similarity. 

Overlaps 

One final set of considerations is about how these defences all work together. There are a number of points to untangle because all four defences can be marshalled at the same time, but some seem to overlap. 

The most complex overlap is between the Wouters rule and the Article 101(3). Some points are clear: Wouters does not apply when the decision of the association is found to restrict competition by object. (I disagree, but this is now the law.) Wouters only applies when there is a decision of an association of undertakings and not when there is another form of agreement. But this leaves a wide field of overlap and here the AG says something worth noting at footnote 67 in RCC Sports: if ‘such rules aim at ensuring that agents act in the best interest of their clients, avoiding conflicts of interest, those rules might be justified under the Meca-Medina case-law or exempted under Article 101(3) TFEU.’  If we consider this passage together with the AG’s reading the notion of progress widely, it seems that according to the AG every positive effect that falls under the Wouters rule could also be discussed under Article 101(3). If this were true, then the Wouters doctrine would become largely irrelevant as it is just applicable to a subset of cases, all of which can also be examined using Article 101(3). Of course engaging Article 101(3) is more tricky because it has an additional requirement – users must secure a fair share of the benefits.  

To a certain extent, the AG is forced into this move by the ECJ’s earlier mistake of stating that the Wouters defence does not apply when there is a restriction by object. This could lead to the paradoxical effect where a decision of an association which  is found to both restrict competition by object and restrict the free movement of services could be justified under the internal market rules because it pursues a public interest, but could not be justified for infringing competition law under the public interest test in Wouters. A wider scope for Article 101(3) is the only way to save the agreement and to treat the justifications offered consistently. 

Two more elegant solutions could have been considered: (1) to say that the Wouters rule applies to public interest justifications, while Article 101(3) TFEU is limited to economic efficiency justifications. (2) to say that Wouters is wrongly decided and to expand Article 101(3). In fact, one might consider that Wouters is a judgment relevant in the pre-modernization days when national courts could not decide that Article 101(3) does not apply, this being the prerogative of the Commission when agreements are notified to it. In that epoch, Wouters solved a procedural tangle. Now that courts (but not NCAs) may decide that Article 101(3) applies, the need for an alternative defence has disappeared.  

A second overlap is when the decision by a dominant association is covered by both Articles 101 and 102. Here the AG Opinion confirms the trend in the case-law by which the justifications that apply to one also apply to the other. The case-law has constructed a de facto Article 102(3) and the AG indicates that the Wouters defence (to the extent that this is different) also applies to the conduct of dominant firms. This is logical because the two prohibitions seek to obtain the same objective.  

A third overlap is when conduct is challenged under both competition law and internal market law. In both Wouters and RCC Sports the challenge was brought under both sets of rules. The AG brings some symmetry to the defences by indicating that under internal market law one applies a balancing test as well, even if this is phrased somewhat differently: in Wouters one must show that the agreement does not eliminate competition, while under internal market there should be a fair balance between the two competing interests. A fair balance entails that the economic freedom of agents is not completely eliminated. 

Conclusions 

The judgment is important for the shape of competition law generally, and not only for the application of competition law to private rule-making by sports bodies. Many might see the AG’s Opinions as opening a pandora’s box of discretion and politics as nearly any public interest counts under either Article 101(3) and/or Wouters. To this response, two replies: first the EU system is designed to take non-competition considerations into account and generations of scholars have pointed this out. Second, the AG recalls that these justifications require evidence not just words.  

The real problem, in my view, is the inelegance by which the theories of harm and the justifications are set up. Suppose you want to challenge anti-competitive regulations: using internal market law seems the smoothest way to go as there is no need to show economic effects. Then justifications require a showing by the defendant that they are protecting a public interest in a proportionate way. But using competition law requires more of a claimant in terms of evidence and we still have some way to go to rationalise theories of harm. Concomitantly, this route provides a confused menu of grounds on which the defendant may rely to justify such conduct, some of which match with the public interest defence in free movement law (Wouters), some which provide a more powerful justification (Albany International where even elimination of competition may be allowed) and some which add further requirements (Art 101(3) with the consumer pass-on test). In my view, this is unnecessarily byzantine and likely to lead to errors and inconsistencies. Perhaps the AG should have advised the Court to start from scratch and build a coherent approach rather than trying to make the case-law fit together in ways that are not always convincing. 

*Following the ASCOLA Declaration, the author has no disclosures to make.

*See further the coverage on this blog of the previous antitrust sports cases (Superleague, ISU, Royal Antwerp): e.g. here, here, here and here.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

PE sponsors are the most frequent notifying parties under the FSR

According to the European Commission’s (EC) FSR registry, there have been 179 FSR filings to date, with the EC currently receiving an average of seven filings per month. Roughly a third of all notifications have been submitted by PE sponsors, making them the single largest category of notifying parties under the FSR regime. FSR filings from PE sponsors span a broad range of sectors, with the most active areas being financial services, consumer goods, construction, energy and industrial equipment.

While the FSR regime only applies to large M&A transactions with a strong EU nexus, PE sponsors often trigger the regime due to the group-wide nature of the financial contribution threshold. Specifically, an FSR filing is required if:

• The target generated at least €500 million EU turnover in the financial year prior to the transaction; and
• The acquirer and the target combined received more than €50 million in foreign financial contributions (FFCs) from non-EU countries over the three years prior to signing the agreement (of acquiring control).

For PE sponsors, this €50 million FFC threshold is frequently met because:

• The test applies at the global group level, including all funds managed by the same GP, and all of the funds’ controlled portfolio companies and JVs;
• Capital raised from Limited Partners (LPs) with non-EU sovereign links may qualify as FFCs, even if they are only passive investors.

The EC is increasingly attuned to the nature of PE structures, particularly around how capital is raised, how funds are managed, and how foreign-state involvement manifests. As a result, the review process is gradually becoming more consistent and predictable — particularly in areas like the assessment of fund-by-fund exemptions and LP disclosures.

The fund-by-fund exemption

One of the most impactful FSR simplifications for PE sponsors is the “fund-by-fund” exemption. It allows them to limit the scope of their FSR disclosure to only the fund(s) acquiring the target (the “Acquiring Fund(s)”), excluding other funds under common management by the same GP, if:

• There is a sufficiently distinct LP base (typically, if 50% or more of the capital / profit interests are held by different investors); and
• There are limited or no commercial transactions between the acquiring and non-acquiring funds or their portfolio companies.

To benefit from the exemption, the EC typically expects:

• An anonymised breakdown of LP commitments (and profit entitlements, where applicable) across the fund structure;
• Confirmation of limited cross-fund dealings, with any inter-fund transactions disclosed and confirmed to be on arm’s-length terms;
• The GP or fund manager is subject to the EU AIFMD (EU Directive 2011/61/EU) or a comparable third-country regime (e.g. in the UK or U.S.).

This exemption is a critical tool to reduce the complexity, information gathering burden and cost of FSR filings. That said, funds should ensure internal compliance tracking to evidence fund separateness and maintain documentation of arm’s-length inter-fund arrangements.

Scope of disclosure for the acquiring fund(s)

The EC takes a close interest in the deal financing and the composition of the LP base. For the Acquiring Fund(s), this includes:

• Anonymised LP list detailing commitments and corresponding capital shares: While identities of non-EU State-linked LPs (e.g. sovereign wealth funds or SOEs) do not need to be disclosed, the presence of such investors must be mentioned, as their contributions may qualify as FFCs.
• Investment terms: The EC assesses whether LPs invest on pari passu terms. Preferential rights (e.g. elevated profit shares or special governance rights) for non-EU State-linked LPs may trigger further scrutiny.
• Governance rights: The EC distinguishes between passive LPs (typical in most PE structures), investors who have meaningful governance rights, such as veto rights, or other powers allowing them to influence investment decisions, and co-investors who invest directly in the target. Co-investors are generally subject to greater interest of the EC, especially if their contributions are earmarked for the deal or where they hold different governance or economic rights to the other investors.
• FFC disclosure: For the Acquiring Fund(s), this includes FFCs only received directly or by controlled portfolio companies or JVs.[1] The breadth of this disclosure obligation will depend on the fund’s maturity and the geographic scope of its portfolio. Where more than one fund is involved, more information will need to be disclosed. Older funds with a larger number of diverse portfolio companies will typically have more FFCs to disclose, while more recently formed funds with fewer portfolio companies will face a lighter disclosure burden. Gathering this information could take time (particularly for PE funds with significant operations outside the EU), but experienced FSR counsel can help navigate this process efficiently.

Waivers

Given the complexity of PE structures, waiver requests are common and PE sponsors account for the majority of them. The EC may grant waivers where:

• The requested information is not reasonably available, e.g. for portfolio companies sold more than three years ago; or
• Certain FFC disclosures are not relevant to the competitive assessment, especially where there are no overlaps in the deal.

In our experience, the EC is pragmatic in handling waiver requests, particularly when parties engage early, and the waiver rationale is clearly explained.

Push for simplification

In the typical LP funding structure, all investors are generally passive and invest on pari passu terms, which makes the FSR filings of the PE funds largely unproblematic. As the EC has become more familiar with PE structures and investment dynamics, it is asking fewer questions than it did at the start of the regime. Pre-notification can also move quickly (as short as one month) where parties front-load the process and come prepared.

In light of the above, the EC has publicly discussed the idea of introducing a simplified FSR filing process for PE funds— whether through a streamlined filing process or an ad hoc simplified procedure for PE deals (given the number of transactions caught, the reporting burden, and the fact that most FFCs relate to passive LP investments and rarely raise substantive concerns). That said, any meaningful changes will likely depend on the EC’s formal review of the FSR, expected to begin in June 2026, which will include a report on enforcement and any proposed changes to revise the FSR thresholds.

While the FSR has introduced new complexity for large-cap PE transactions in Europe, the regime is stabilising. The EC has shown flexibility and a willingness to adapt to market realities — especially for PE funds, which are among the most active dealmakers and most impacted by the regime’s mechanics.

****

Anastasios Tsochatzidis (White & Case, Trainee, Brussels) contributed to the development of this post.

[1]     The concept of “control” under the FSR is the same as under the EU Merger Regulation. Pursuant to Article 3(2) of the EU Merger Regulation (replicated in Article 20 of the FSR), “Control shall be constituted by rights, contracts or any other means which, either separately or in combination and having regard to the considerations of fact or law involved, confer the possibility of exercising decisive influence on an undertaking, in particular by: (a) ownership or the right to use all or part of the assets of an undertaking; (b) rights or contracts which confer decisive influence on the composition, voting or decisions of the organs of an undertaking”.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

Ukraine’s decade-long selective distribution practices in the pharmaceutical market are currently facing unparalleled antitrust scrutiny. This follows the introduction of controversial new regulation that reflect a growing trend: substituting timely, targeted market failure interventions and adequate enforcement with legislative restrictions that are economically unsound and severely undermine decades of the established commercial strategies.

The Current Market Context

The Ukrainian wholesale pharmaceutical distribution market has historically evolved into a duopoly dominated by two major players — BaDM and Optima. Benefiting from political support and substantial investments in developing nationwide distribution networks, these two companies have maintained an undisputed leadership position, collectively controlling over 85% of the market. Numerous and repeated attempts by the government to dismantle this distribution monopoly have yielded no tangible results.

The Antimonopoly Committee of Ukraine (AMCU) has been conducting a year-long investigation into potential abuse of dominance by these distributors. Despite recent public promises from the AMCU leadership to finalize the investigation and adopt a decision soon, it seems unlikely that the case will lead to anything beyond financial penalties. Meaningful structural remedies that could genuinely open the market to broader competition are even less probable. The core problem persists: no other player — particularly under the challenging conditions of wartime — is willing or is able to invest significant resources in developing their own nationwide pharmaceutical distribution infrastructure to challenge the entrenched duopoly.

At the same time, pharmaceutical manufacturers and importers operating in Ukraine have developed internal policies over the years which set a number of qualitative and quantitative criteria for selecting distribution partners to ensure compliance with comprehensive quality and safety requirements — criteria that, in practice, only BaDM and Optima are able to meet.

Regulatory Response

In response to this market failure, the Ukrainian government earlier this year opted for a legislative solution, introducing imperative legal provisions into pharmaceutical legislation that drastically alter market rules — particularly those governing the relationship between manufacturers, importers, and distributors.

The central new requirement limits the volume of any given medicinal product that can be sold to a single distributor to no more than 20% of total annual sales for that product. In addition, manufacturers and importers must offer identical commercial conditions — including price, delivery terms, and payment deadlines — to all distributors. This restriction does not apply to the so-called original (innovative) medicinal products, the list of which is maintained by a specialized state agency.

AMCU’s Advocacy Position

In early May, following its advocacy mandate, the AMCU issued mandatory recommendations to nearly all market participants (160 companies, including pharmaceutical manufacturers and importers). The core message was clear: manufacturers should avoid setting distributor selection criteria that unjustifiably favor certain distributors (i.e. BaDM and Optima) which may lead to a distortion of competition in the pharmaceutical distribution market.

Notably, these recommendations were also addressed to manufacturers of original medicinal products, even though the “20% quota” rule does not formally apply to them. This move reveals several important insights:

• The AMCU’s position appears to be that it is not strictly bound by sector-specific pharmaceutical legislation. While it acknowledges the new regulations, the AMCU operates under a broader competition protection mandate that covers the entire pharmaceutical distribution market.
• The new “20% quota” restriction is likely to serve as a reference point for the AMCU. However, based on its recommendations, it is clear that all market players are expected to comply with economically justified and competition-friendly selective distribution practices.
• Even manufacturers of original medicinal products, formally exempt from the “20% quota” requirement, will conceivably need to review and adapt their distribution policies so as to ensure their selective distribution models do not trigger AMCU concerns under the new regulatory environment and to align with evolving regulatory expectations.
• The fundamental problem — the absence of new entrants capable of disrupting the BaDM–Optima duopoly — have few chances to be resolved. However, manufacturers and importers who continue operating under outdated models without adjusting their selective distribution systems and commercial policies may find themselves under AMCU scrutiny.
• Consequently, the pharmaceutical sector should anticipate increased investigations and antitrust cases initiated by the AMCU. Yet these efforts, while having a prospect to improve formal compliance, are barely feasible to correct the underlying market failure or meaningfully intensify competition in pharmaceutical distribution.

Between Duopoly and Regulation

Ukraine’s experience should serve as a cautionary tale. Introducing rigid, administratively imposed restrictions in the absence of real market alternatives rarely leads to sustainable competitive outcomes. What makes matters worse, it risks destabilizing the well-established commercial ecosystems and undermining operational efficiency, particularly in a market as strategically important as pharmaceuticals.

Presumably, a better strategy would have been to focus on enhancing competitive conditions through targeted enforcement, facilitating market entry opportunities, and addressing specific abuses of dominance where proven — not to enforce blanket rules that may ultimately harm both market performance and consumer welfare, while the broader question of fostering genuine competition remains unresolved.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

In this blog post, we provide a summary of the year’s key developments. We begin with fine statistics and enforcement trends, highlight priority sectors targeted by the regulator, and review merger control activity. We then examine enforcement in areas such as bid rigging, unfair competition, and informational violations; spotlight the first-ever application of the leniency procedure; and summarize the AMCU’s efforts in state aid control and competition law reform. Finally, we outline the AMCU’s enforcement priorities and provide an outlook for 2025.

Fine Statistics

Throughout 2024, the regulator issued 1,146 decisions on violations of Ukrainian competition law, imposing fines totaling just over UAH 1 bln. (approx. EUR 23 mln.). Of the UAH 1 billion in fines imposed by the regulator in 2024, nearly 90% stemmed from two types of violations: anticompetitive concerted practices and gun jumping. Notably, bid rigging accounted for 100% of the total fines imposed for anticompetitive concerted practices. The AMCU also investigated 47 gun-jumping cases, representing 10% of the total fines imposed. Violations related to unfair competition, informational violations, and abuse of dominance accounted for 5.2%, 2.9%, and 2.8% of the total fines, respectively.

Figure 1. Breakdown of the AMCU fines by Type of Violation

Sectors Targeted by the AMCU

Throughout 2024, the regulator enforced Ukrainian competition law across a broad range of industries, namely:

Figure 2. Sectors at the AMCU’s Spotlight in 2024

Merger Control

The AMCU was busy with merger control reviews in 2024, having reviewed 496 notifications, of which:

• 364 were processed and resulted in merger clearances;
• 131 were either rejected due to non-compliance with formal requirements or withdrawn by the parties on their own initiative. It is worth noting, however, that most of the rejected notifications were subsequently rectified and resubmitted by the parties, ultimately receiving merger clearance and contributing to the total of 364 approved mergers [1];
• 1 case was closed without a decision on the merits.

The majority of clearances (306 out of 364) were granted in Phase I as the respective mergers posed no threat to domestic competition. Competition concerns, however, were identified in 11 cases, which were ultimately cleared in Phase II following an in-depth review by the AMCU – 4 of these were subject to remedies. The remaining 47 clearances were issued in gun-jumping cases, where transactions were completed without prior AMCU’s clearance, but the parties voluntarily submitted their filings to the regulator post-closing.

Also, in 2024, the AMCU issued 51 preliminary conclusions on mergers. These are issued in response to specific notifications and reflect the AMCU’s initial assessment of whether a transaction requires merger clearance and whether such clearance is likely to be granted.

Figure 3. 2024 Merger Control Statistics

Notable Mergers

The largest transactions by value cleared in Phase I by the AMCU in 2024 included:

• $8.5 billion merger of Reliance Industries’ and Walt Disney’s Indian media assets;
• $4 billion acquisition of Broadcom’s End-User Computing Division by KKR;
• $3.6 billion merger of Berry Global’s Health, Hygiene and Specialties Global Nonwovens and Films business with Glatfelter.

Out of deals having appreciable nexus in Ukraine, the following are worth mentioning:

• Merger of Bunge and Viterra. The AMCU’s assessment focused on two markets: agricultural products and the transshipment of vegetable oil at the Mykolaiv port. Regarding the first market, the AMCU concluded that the parties’ combined market shares were moderate and the market itself was highly competitive. As for the second market, the regulator noted that the importance of the Mykolaiv port had declined following the beginning of the full-scale war in Ukraine in February 2022. This led to a shift in demand to other ports, along with a decrease in the volumes of oil production and transshipment. The AMCU concluded that the transaction would not result in monopolization or a restriction of domestic competition and, therefore, granted the clearance.
• Acquisition by MSC of Hamburger Hafen und Logistik AG and the Port of Hamburg.
  The AMCU concluded that the transaction could adversely affect competition in the container terminal of the Odesa seaport. To mitigate this risk, the regulator cleared the transaction subject to certain behavioral remedies imposed on the buyer, in particular, an obligation not to unjustifiably restrict third-party access to container terminal services and terminal’s infrastructure. Access must be granted on fair market terms and with at least 30% of the terminal’s annual available design capacity allocated accordingly.
• Acquisition by CRH of Dyckerhoff Cement Ukraine. The parties were close competitors with high market shares in the Ukrainian market for gray Portland cement. To prevent the monopolization of the market, the transaction was cleared subject to certain structural remedies, including:
  (1) the divestiture of 25-28% of shares in Dyckerhoff Cement Ukraine (Dyckerhoff), along with certain veto rights, to an independent third party; (2) the appointment of only independent executives at Dyckerhoff; (3) the prohibition to unjustifiably refuse to supply cement to third parties, as well as the obligation to maintain existing production capacities; (4) the application of market-based pricing and fair contract terms to all cement buyers; (5) the prohibition to unjustifiably restrict third-party access to the cement market; and (6) the prohibition on clinker exports that would prevent fulfillment of third-party cement orders.

Gun Jumping

Local Transactions

In 2024, the AMCU imposed fines totaling UAH 101.5 mln. (approx. EUR 2.3 mln.) for gun-jumping violations. More than a half of that amount was imposed in the following local transactions:

• UAH 37.3 mln. (approx. EUR 0.84 mln.) fine for the acquisition of control by PJSC “Kyiv Cardboard and Paper Mill” over LLC “Autospetstrans-Kyiv Cardboard and Paper Mill,” which took place back in 2017;
• UAH 21.3 mln. (approx. EUR 0.48 mln.) fine for the acquisition by LLC “Pulp Mill Print” of certain assets and entity of Blitz-Inform group, which took place back in 2019.

The regulator identified that the buyers in these transactions (PJSC “Kyiv Cardboard and Paper Mill” and LLC “Pulp Mill Print”) belong to the same corporate group. Repeated violations by entities within the same group may be considered an aggravating factor by the regulator and could contribute to the imposition of higher fines.

Foreign-To-Foreign Transactions

Additionally, in 2024, the AMCU imposed fines for closing foreign-to-foreign transactions without the regulator’s clearance:

• UAH 4.5 mln. ( EUR 0.11 mln.) fine for the creation of a joint venture between Electricite de France, Nebras Power, Sojitz Corporation and Kyuden International Corporation;
• UAH 2.5 mln. ( EUR 0.06 mln.) fine for the acquisition by Sika of control over MBCC;
• UAH 0.47 mln. ( EUR 0.01 mln.) fine for the acquisition by YILFERT Holding of control over Rosier SA; and
• UAH 0.05 mln. ( EUR 0.001 mln.) fine for the acquisition by Cheplapharm of control over certain pharmaceutical assets. [2]

Bid Rigging

In 2024, the AMCU continued its traditional focus on identifying bid-rigging violations – i.e., the distortion of results of bids, auctions, tenders, and public procurements – which constitute one of the types of anticompetitive concerted actions under Ukrainian competition law. Last year, the regulator issued 2,012 decisions related to anticompetitive concerted actions, 100% of which being bid-rigging violations. The highest fine imposed for bid rigging during the year totaled UAH 106 mln. (approx. EUR 2.4 mln.).

It is worth mentioning that the AMCU has developed strong expertise in tackling this type of violation. In Ukraine, bid rigging is considered a per se violation, and thus the regulator is not required to prove a negative effect on domestic competition. Furthermore, fines for bid rigging may reach up to 10% of the annual turnover per violation; however, the most adverse consequence is not the fine, but rather the prohibition from participating in public procurements for the next three years.

Unfair Competition

In 2024, the AMCU identified 58 violations of the Law of Ukraine “On Protection Against Unfair Competition” (Unfair Competition Law), with total fines amounting to UAH 53.3 mln. (approx. EUR 1.2 mln.). The highest fine imposed by the AMCU for unfair competition last year was UAH 17.7 mln. (approx. EUR 0.4 mln). Traditionally, the most frequent violation of the Unfair Competition Law in the country has been the dissemination of misleading information.

In 2024, the AMCU focused on identifying the dissemination of misleading information across three markets: mineral waters, dietary supplements, and cosmetic products. It is worth noting that the threshold for proving misleading information in Ukraine is relatively low, as the applicable legal test in Ukraine is based on the perspective of an average, unskilled consumer – someone who is presumed not to verify the accuracy of commercial claims and who relies primarily on their overall first impression.

Informational Violations

In 2024, the AMCU continued its active efforts to combat so-called informational violations – i.e., providing false, inaccurate, or incomplete information to the regulator in response to its request, or failing to submit the requested information altogether. The regulator reviewed 312 cases of informational violations in 2024, with the most common being the failure to submit the requested information.

Nord Stream 2 Project–Related Case

The case stemmed from the AMCU’s investigation into alleged anti-competitive agreements related to the Nord Stream 2 project (Project). The AMCU launched an investigation back in 2021, suspecting a gun-jumping violation committed by the Project participants – specifically, the failure to obtain prior concerted practice clearance from the regulator. As part of the investigation, the AMCU sent the requests to the Ukrainian subsidiaries of: (1) Engie Energy Management Holding Switzerland AG (Engie), (2) Shell Exploration and Production (LXXI) B.V. (Shell), and (3) Wintershall Nederland Transport and Trading B.V. (Wintershall) requesting to confirm or deny conclusion of any agreements related to the Project.

According to the AMCU, none of the entities submitted information to the regulator, preventing the latter from exercising its discretionary powers – in case at hand, determining whether the participants in the Project were required to obtain prior clearance. As a result, the AMCU imposed fines on the Ukrainian subsidiaries of Engie, Shell, and Wintershall. Shell’s subsidiary received a fine of UAH 43 mln. (approx. EUR 0.9 mln.) – the largest fine ever imposed by the AMCU for an informational violation. All three entities challenged the AMCU’s decision in Ukrainian courts.

In March 2025, Wintershall’s Ukrainian subsidiary won the case before the Ukrainian Supreme Court. The central question in this case was whether Ukrainian entities belonging to foreign corporate groups can be held liable for failing to provide the regulator with the requested information that is in the possession of another entity within the same corporate group. The Supreme Court ruled that holding a Ukrainian entity liable for not providing information held by a foreign affiliate would violate the principle of fairness in legal liability. Furthermore, it emphasized that corporate affiliation alone does not justify extending the AMCU’s enforcement powers beyond Ukraine’s jurisdiction.

AMCU Started Targeting Digital Markets

Jumping ahead, digital markets have been identified as the AMCU’s top enforcement priority for 2025. In fact, the regulator began moving in this direction as early as 2024, issuing recommendations to two tech-related companies operating in different areas of the digital economy.

Kyiv Digital

In early 2024, the AMCU issued recommendations to the municipal company behind the Kyiv Digital app – a multifunctional platform offering various services, including payments for transport ticketing and parking fines. The AMCU expressed concerns that users faced commission fees of up to 10%, with no access to alternative payment service providers (PSPs) that could offer more competitive rates. To address these concerns, the regulator issued recommendations to the municipal entity administering the app to promote competition by enabling the integration of additional PSPs on transparent, clear, and non-discriminatory terms.

BlaBlaCar

Later in the year, the AMCU turned its attention to BlaBlaCar, identifying the platform as likely dominant in the narrowly defined market of ”carpooling services via online platforms in Ukraine”, citing its market leadership and lack of substantial competition from rivals. The regulator alleged that BlaBlaCar applied unjustified service fees to passengers and maintained inconsistent refund policies when passengers canceled rides. The AMCU recommended that the company adopt a transparent methodology for fee calculation and implement a fair and non-discriminatory refund policy.

Legislative Developments and EU Integration

Last year, the AMCU remained actively engaged in supporting Ukraine’s EU integration efforts. As part of the second phase of the Ukrainian competition law reform, the regulator developed legislative proposals aimed at strengthening its institutional capacity and providing more effective tools for enforcing competition law. Notably, the proposals introduce administrative liability for companies and government officials who obstruct dawn raids, fail to submit requested information to the regulator, or ignore official summonses to appear and provide explanations.

On the international front, the AMCU took significant steps toward obtaining associate membership in the OECD Competition Committee, further aligning Ukraine’s competition framework with global best practices, including implementation of the OECD Recommendations on Competitive Neutrality. Throughout the year, the regulator also cooperated with the European Commission on competition policy, holding bilateral meetings as part of the official screening process to assess the alignment of Ukrainian legislation with EU law.

First-Ever Leniency Case 

2024 marked a milestone for Ukraine’s antitrust enforcement with the first-ever application of the updated leniency procedure by the AMCU. Although leniency provisions had been part of Ukraine’s competition law since 2002, their effectiveness had long been hindered by the lack of clear procedural guidance. A significant shift occurred with the entry into force of competition law amendments on 1 January 2024, accompanied by the adoption of the respective procedure by the AMCU.

Ukraine’s updated leniency procedure brings greater clarity and alignment with EU standards. It offers full immunity from liability to the first applicant, while subsequent applicants may receive fine reductions of up to 50%, 30%, or 20%, respectively. Applications can be submitted even after an investigation has started, as long as they are filed before the AMCU issues its preliminary conclusions.

In December 2024, the AMCU applied the updated leniency procedure in a bid-rigging case involving two public procurements. One company received full immunity after providing information about the violation to the regulator, submitting key evidence and fully cooperating. The second company, which did not cooperate, was fined 2% (the maximum statutory fine of up to 10%) of its annual turnover for each of the two bid-rigging violations committed.

State Aid

To recall, Ukraine introduced a state aid regime in alignment with EU acquis under the EU-Ukraine Association Agreement in 2014. The regulatory framework, effective from 2 August 2017, is based on the law of Ukraine “On State Aid to Undertakings” (State Aid Law) and secondary legislation.

Following the full-scale invasion by Russia, Ukraine introduced martial law, leading to a temporary suspension of state aid control, effective from March 2022. Since then, state aid grantors have been exempted from the obligation to notify the AMCU of new or amended aid measures – these measures are automatically deemed compatible with domestic competition rules.

In 2024, the AMCU prepared a draft law aimed at introducing liability and penalties for state aid grantors, particularly those who grant unlawful or incompatible state aid. In line with its obligations under the EU-Ukraine Association Agreement, the AMCU conducted a partial inventory of existing schemes and compiled a list of potential state aid measures requiring compliance with the State Aid Law. Grantors must align these schemes with the law and its compatibility criteria. Non-compliant schemes will be terminated.

AMCU’s Top Priorities for 2025

In its 2024 report, the AMCU highlighted the following markets and industries its top enforcement priorities for 2025:

Figure 4. AMCU’s Top Priorities for 2025

The AMCU also highlighted in its report that identifying bid-rigging violations in the following sectors will remain a top priority in the upcoming year: defense, healthcare, and construction. Another key priority for the AMCU in 2025 will be enhancing its institutional capacity, including harmonizing Ukrainian competition law with EU standards, continuing cooperation with the OECD, and further strengthening the AMCU’s powers in bid-rigging investigations.

Outlook for 2025

Despite the ongoing challenges of wartime, we expect the AMCU to remain fully operational and increasingly active in 2025. The regulator is likely to issue a greater number of decisions, impose higher fines, and commence new investigations and sector inquiries.

We do not anticipate a decline in the average number of merger control notifications reviewed by the AMCU. Over the past five years, this number has consistently ranged between 450 and 550. Given the AMCU’s growing focus on identifying gun-jumping violations, we strongly recommend that foreign counsel assess carefully whether a transaction is notifiable and ensure timely notification when required.

Finally, we expect the AMCU’s increasing interest in digital markets to translate into more assertive enforcement. The regulator has clearly identified digital platforms as its top enforcement priority for 2025. Accordingly, we anticipate a rise in investigations involving tech companies operating in a digital economy.

[1] This means that if a transaction involves the acquisition of multiple direct targets – rather than a single holding company consolidating them – there may be multiple separate filings for the same overall deal. As a result, the statistics may be distorted, and the actual number of distinct deals could be lower than the number of filings suggests.

[2] The exceptionally low fine was because the transaction qualified for a special procedure available during the initial months of the war (February–June 2022), under which only nominal fines were imposed in cases where the parties filed and closed the transaction before obtaining the clearance.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

EVP Ribera’s Mission Letter raised questions about how broader European Union (EU) policy objectives could or should be integrated into the Commission’s assessment of transactions under the EU Merger Regulation (EUMR).  In comments to the European Parliament, EVP Ribera said that “EU merger control must continue to evolve to capture contemporary needs and dynamics like globalization, digitalization, sustainability, innovation and resilience.”  Launching the Consultation, EVP Ribera described it as a “comprehensive and ambitious review . . . to account for disruptive changes in our societies and our economies over the past 20 years, such as digitalisation, and enable us to ensure that innovation, resilience, and the investment intensity of competition are given adequate weight in light of the European economy’s acute needs.”  She argued that “only by evolving . . .  can [we] ensure that our merger control policy continues to serve people, drive innovation, and strengthen Europe’s resilience and leadership.”

EVP Ribera’s Mission Letter and comments on the Consultation, as well as the text of the Consultation itself, suggest that the future Merger Guidelines will reflect a thorough-going review of the basic principles of EU merger control, not merely a technical update to reflect EU Court judgments and the Commission’s decisional practice in the years since the Merger Guidelines’ adoption.

The Consultation.  The Consultation is divided into two parts, a General Questionnaire and a targeted consultation with an In-depth Questionnaire covering “topics that are key for the EU economy, namely competitiveness and resilience, market power, innovation, decarbonisation, digitalisation, efficiencies, defence and labour considerations.” Alongside the targeted consultation, the Commission published seven papers that will be the basis for broader engagement with stakeholders, including through dedicated events and workshops. The Commission will also commission an economic study on the dynamic effects of mergers.

The Consultation refers to the Commission’s July 2024 Political Guidelines and EVP Ribera’s mission letter, as well as the January 2025 Competitiveness Compass.  The Consultation also draws inspiration from the September 2024 Draghi Report on European competitiveness, which among other things called for a new approach to EU merger policy allowing European companies to achieve greater scale and the introduction of a new “innovation defence”.

The Consultation is open until September 3, 2025.  The Commission will then publish the feedback it receives and its own evaluation, followed by a consultation on draft revised Merger Guidelines.  The Commission will also conduct and publish an impact assessment. The Commission aims to adopt the final revised Merger Guidelines in late 2027 but can be expected to begin implementing any policy changes stemming from the Consultation well in advance.

The General Questionnaire.  The General Questionnaire observes that, “in the respectively twenty-one and sixteen years since the adoption of the [Merger] Guidelines there have been significant market trends and developments that have changed the dynamics of competition. . . .In light of these factors, which apply equally to both the [HMG and NHMG], the Commission is proposing to revise both sets of guidelines in a holistic exercise. The goal is to ensure the Guidelines are up-to-date in order to allow the Commission to continue to protect competition under the [EUMR] in evolving market realities, while not intervening in transactions that do not harm competition. In addition, the revised merger guidelines should provide increased transparency and predictability to the business community as to how the Commission assesses mergers today.”

The General Questionnaire accordingly asks broad questions, including whether the Merger Guidelines have “allowed the Commission to identify correctly the transactions that significantly impede effective competition in the internal market;” “contributed to promoting competition;” and provided “correct, clear and comprehensive guidance on merger assessment” and “legal certainty and transparency”.  The General Questionnaire also asks about the costs and benefits of having Merger Guidelines; the objectives the Merger Guidelines do or should pursue; possible inconsistencies or contradictions; and whether the Merger Guidelines contribute to more consistent enforcement, as well as soliciting suggestions for simplification and cost-reduction.

Although the Commission commonly asks such threshold questions about whether antitrust policy documents should be renewed, in practice there is little or no doubt that the Commission plans to issue revised Merger Guidelines.  On the other hand, the General Questionnaire asks several questions that may point to changes in the structure and content of the future guidelines.  For example, the General Questionnaire asks whether “the distinction between effects of horizontal and non-horizontal mergers [is] still relevant” and whether it would be preferable to have separate guidelines on horizontal and nonhorizontal mergers or a single document. The General Questionnaire also asks whether and if so how future guidelines should take account of sectoral regulations (e.g. telecommunications, energy) and particular features of certain sectors (e.g., longer investment cycles, innovation intensity).

The General Questionnaire also seeks input on the Merger Guidelines’ treatment of factors used to assess market power, such as market shares, concentration level, barriers to entry or expansion, and diversion ratios, whether certain aspects are unclear or outdated and whether other metrics should be included.  The Commission further seeks input on the assessment of coordination and foreclosure risks and anti-competitive effects that may stem from transactions that do not create or strengthen a dominant position.

The General Questionnaire touches on themes addressed more fully in the In-depth Questionnaire. These include “competitiveness” (including the benefits of increased scale, security of supply, resilience of the EU economy and increased innovation and investment).  More specifically, the General Questionnaire seeks input on potential harms and benefits of consolidation in global strategic sectors, digital and deep technology innovation, and clean and resource efficient technologies and biotechnologies (e.g., IoT, cloud, quantum, telecom, data, advanced connectivity, cybersecurity, and/or AI).  Similarly, the General Questionnaire calls for input on the treatment of innovation and other dynamic elements; sustainability and clean and resource-efficient technologies; digitalization; efficiencies; public policy, defense and security; and labor market considerations.

The In-depth Questionnaire.  The In-depth Questionnaire “focusses on in depth and technical parameters related to EU merger control” on the following seven topics: competitiveness and resilience; assessing market power using structural features and other market indicators; innovation and other dynamic elements in merger control; sustainability and clean technologies; digitalization; efficiencies; and public policy, security and labor market considerations.  These questions apparently represent the Commission’s efforts to collect evidence that can be used to implement the Mission Letter’s broader mandate to modernize EU competition law.

For each topic, the In-depth Questionnaire provides a brief introduction and technical background, followed by specific questions. .In relation to competitiveness and resilience (Topic A), the In-depth Questionnaire calls for reflection on whether EU merger control must be adapted to support start-ups, scale-ups, and medium-sized companies to scale up in global markets, while safeguarding a level playing field. The In-depth Questionnaire identifies four specific topics for further investigation: scaling up; resilience and value chains; enhancing investment and innovation; and globalization.

The In-depth Questionnaire notes that productivity tends to increase scale, and increasing scale through mergers and acquisitions may help firms become more productive. The acquisition of existing businesses may also be a means for a company to expand into other Member States or increase its global outreach to compete with large global rivals. On the other hand, market power resulting from mergers can lead to price increases, diminished quality or innovation, and a reduced number of suitable suppliers, all of which can negatively impact the competitiveness of other businesses.

Mergers may also have a negative or positive impact on the EU’s resilience in the face of global shocks and the need for a diverse, competitive supply base (e.g. for critical raw materials and other inputs required for the green and digital transitions). On the one hand, mergers can secure companies’ access to inputs they need to compete, including through the integration of activities at different levels of the value chain, and integration of competitive EU suppliers may reduce dependencies on external sources. On the other hand, mergers may result in less competitively priced inputs, less innovative or lower quality products or a reduced number of suitable suppliers, with negative effects on companies’ competitiveness and resilience not only in Europe but also in global markets. Having a variety of businesses active in the EU Single Market can support firms’ ability to multi-source and to be dynamic and resilient to shocks. By contrast, less competition risks making an economy “brittle” and thus less resilient.

Scale resulting from M&A transactions can also impact incentives for investment and innovation.  Scale might provide companies with benefits such as lower costs, better access to capital markets or R&D&I capabilities that increase their ability to invest and innovate. At the same time, company size does not typically reflect the ability to invest and innovate, as many of the most innovative firms in sectors such as pharma, biotechnology, digital or high-tech are small and medium-sized enterprises.  While scaling up companies with disruptive technologies can help disseminate important innovations across the economy, the acquisition of nascent competitors by large established players to protect their market power (so-called “killer acquisitions”) might harm innovation.

The In-depth Questionnaire observes that the degree of globalization affects the geographic scope of competition in relevant antitrust markets. Competition in the EU may be affected by imports into Europe from other parts of the world, but also by subsidies or other competitive advantages received by market participants outside the EU.

In relation to the assessment of market power using structural features and other market indicators (Topic B), the HMG and NHMG both contain structural indicators relating to market shares and concentration levels that mostly provide guidance on where competition concerns are unlikely to arise (so-called “safe harbors”). With the exception of market shares above 50% in a horizontal merger, they do not offer rules of thumb for when a merger can be presumed to be harmful, since there can be situations where a merger will not harm competition, for instance because the parties are not close competitors, competition in the market is intense, or large market shares may turn out to be only temporary.

In the Commission’s view, the revision of the Merger Guidelines offers a chance to adequately reflect the risks resulting from mergers in a situation of rising levels of concentration and profit margins in EU markets.  One means to achieve this would be the adoption of stricter indicators (or rebuttable presumptions) to identify more easily mergers that are likely to result in a significant impediment to effective competition. In addition, the Commission may set out a more comprehensive framework relying on alternative approaches to assessing market power, and particularly those that emerged in its case practice. For example, capacity shares are already frequently used structural indicators. Further market features of relevance may include diversion ratios, profit margins, the distribution of spare capacities or a firm’s pivotality. Some of these market features may be especially relevant in cases that do not result in the creation or strengthening of a dominant position, or in cases involving highly differentiated markets.

The revised Merger Guidelines may also reflect criteria for the assessment of cases that do not result in the creation or strengthening of a dominant position. For instance, the revised Merger Guidelines may provide further guidance on when merging firms can be considered close competitors or how to identify mergers that would result in the elimination of an important competitive force. In some cases, even if the combined market shares or concentration levels are not particularly high, a merger may still lead to anticompetitive effects by increasing the risk of coordination. Given developments such as algorithmic pricing, the In-depth Questionnaire calls for reflection on whether the framework for the assessment of coordinated effects is still fit for purpose. Similarly, the In-depth Questionnaire calls for reflection on whether the Merger Guidelines’ “ability-incentive-effects” framework for assessing foreclosure risks in non-horizontal mergers should be amended.

In relation to innovation and other dynamic elements in merger control (Topic C), the In-depth Questionnaire notes that mergers can impact innovation competition in both directions – they may increase the ability of the merged firm to innovate but also harm innovation competition and thus incentives to invest in R&D. The effects of mergers on innovation are often more difficult to predict than price effects, so the challenge is to further develop a sufficiently accurate yet administrable framework for assessing dynamic merger effects on innovation.

Similarly, the acquisition of a potential competitor with a promising product in development or notable R&D capabilities can accelerate commercialization of improved products or prevent future competition (e.g. if a merger leads to the discontinuation of a promising product or line of research or increases barriers to entry or expansion). The challenge is to identify circumstances in which the acquisition of a potential competitor may increase or stifle competition, not only in horizontal but also in non-horizontal mergers.

Topic D addresses the role of sustainability and clean technologies. The In-depth Questionnaire notes that merger control has a role to play in allowing procompetitive mergers to support the transition to a clean and sustainable economy, while preventing mergers with negative effects on clean innovation and sustainability goals, for example, where an incumbent acquires a disrupting innovator offering a green product to delay or cannibalize it (“green killer acquisitions”) or a merger reduces incentives to invest and innovate in green products or technologies. Non-horizontal mergers may also have a negative impact, for instance by removing or reducing access to less carbon- or energy-intensive products or services (including key green technologies and materials, such as batteries, renewable components, and recycling infrastructure) that generate less waste or require less raw materials.

On the other hand, mergers may support climate and sustainability objectives and the clean transition and have a positive impact on clean innovation, for example on the deployment of cleaner/greener technologies or manufacturing processes. Mergers can provide the leverage needed to invest in decarbonization, cleaner products and technologies and more energy-efficient solutions and infrastructure. Vertical integration may also enhance the circular use of raw or recycled materials and allow companies to adopt more innovative, efficient and clean resource management across larger segments of the supply chain.

Some mergers may also generate sustainability benefits that could offset negative effects on competition (“green efficiencies”). At the same time, careful assessment is required to avoid greenwashing attempts and ensure that claimed benefits materialize post-merger. Mergers should not make clean products or services, for example, related to renewable energy, sustainable waste management and recycling, resource-efficient (digital) solutions, electric vehicles etc., less affordable or inaccessible to businesses and citizens.

The In-depth Questionnaire notes that the growing interplay between competition, innovation and sustainability considerations across industries and related benefits calls for reflection on merger control’s contribution to European sustainability objectives. Key questions in this regard include the methodology and parameters to be included in the competitive assessment to take due account of sustainability considerations, as well as the quantification and verification of green incentives and efficiencies.

Digitalization (Topic E) has of course been a key feature of markets addressed in EUMR reviews for decades.  The In-depth Questionnaire notes that an extended forward-looking assessment may be required to properly capture the effects of a transaction in such markets, particularly when the merger involves the acquisition of a nascent player or a nascent market. In fast-moving markets, killer acquisitions of complements require careful assessment, because in such markets a complementary product or player of today may very quickly become a substitute.

According to the In-depth Questionnaire, leading companies in the digital and tech sectors commonly seek to acquire complementary businesses or key inputs (e.g., data, technology, user traffic, but also talent, compute capacity and others) with the aim of strengthening their position in core markets. Such a strategy may contribute to increases in innovation. On the other hand, developing or expanding an ecosystem of related products and services may entrench an incumbent’s position, thus making it harder for rivals to enter, expand, or innovate.

This type of business strategy does not easily fit into the traditional distinction between horizontal and non-horizontal mergers, because fewer transactions are purely horizontal, vertical or conglomerate in nature, and the lines between horizontally or non-horizontally linked product markets become increasingly blurred. For instance, in mergers involving companies with activities across several product markets, products often need to interoperate with each other or are offered as part of an ecosystem of related services.

Digital markets also raise questions about how forward-looking merger assessments should be, what kind of future changes should be taken into account, and what facts and evidence should be considered. This is particularly challenging in nascent and fast-moving markets, where historical market shares may say little about future effects on competition.

Finally, certain digital mergers raise privacy and data protection concerns, for instance when a merger leads to the acquisition of data or the combination of datasets. For example, competition to gain customers based on companies’ privacy settings can be considered a non-price parameter of competition, and the acquisition of a target marketing itself as prioritizing customer data protection could reduce consumer choice for privacy-focused services. Privacy concerns can also be relevant to the credibility of (alternative) suppliers, e.g., if customers don’t find it feasible to work with suppliers processing data in non-EU servers.  The question is whether and if so how privacy and data protection objectives enshrined in EU law should be taken into account as parameters of competition.

In relation to efficiencies (Topic F), the In-depth Questionnaire notes that otherwise harmful mergers may result in “efficiencies” that may counteract potential harms to consumers. Mergers can generate cost savings that are passed-on to consumers in the form of lower prices or lead to improved products or services, for example from increased investment and innovation (as opposed to synergies that only result in higher profits). Compared to horizontal mergers, vertical and conglomerate mergers may provide more scope for efficiencies, e.g. in the form of an elimination of double margins or better coordination of marketing efforts.

Efficiencies should be assessed against the EUMR’s legal mandate to protect effective competition and the clarification that any efficiencies should be to the advantage of intermediate and ultimate consumers. The Merger Guidelines specify that cognizable efficiencies must benefit consumers, be merger-specific and be verifiable.  The balancing exercise between harm and efficiencies becomes increasingly complex when there is asymmetry between the alleged anticompetitive effects and benefits.  Another challenge arises from timing differences, as investments usually materialize over a long period of time, whereas anticompetitive effects may materialize immediately.

The question arises which type of evidence or metrics are appropriate for the assessment of efficiency claims and the required likelihood of materialization to accept efficiencies. For example, the assessment of efficiencies concerning improved quality of products or services is typically linked to consumers’ willingness to pay for higher quality, and merging companies may find it difficult to submit reliable and robust evidence in support of the increase in quality.

Finally, efficiencies have to be merger-specific. The Commission must consider whether the same benefits could be achieved in a less harmful way, for example through a cooperation agreement. However, determining the existence and viability of an alternative may not be straightforward. For instance, an alternative option should be realistic, but this may be put into question if an acquirer has already made an unsuccessful attempt at it in the past.

Topic G groups together public policy, security and labor market considerations.  Although these topics are very different, each raises a question about how EU policy objectives beyond competition policy can be integrated into EU merger review.   Although merger control focuses primarily on ensuring that mergers do not harm consumers, vibrant competition also contributes indirectly to other policy objectives. Where companies become too powerful, they may become too-powerful-to-care. Where companies become so large as to be essential, they can become too-big-to-fail, and therefore increasingly difficult to regulate.

The Political Guidelines call for a new era for European Defence and Security, and there have been calls for further consolidation in the EU defense sector. Neither the HMG nor the NHMG include guidance specific to mergers relating to security or defense. Member States may consider legitimate national security interests to be impacted by a merger – and consequently seek to intervene on public security grounds.  However, merger rules may prevent harmful market power in non-European inputs relevant for EU defense. Thus, the In-depth Questionnaire seeks feedback on whether the revised Merger Guidelines should deal with the interaction between Member States’ security and defense interests and the Commission’s competition assessment.  Similarly, the In-depth Questionnaire seeks feedback on how to undertake a potential balancing of interests between defense and competition objectives in cases involving dual-use goods.

Mergers can also impact media plurality. Article 21(4) EUMR allows Member States to “take appropriate measures to protect legitimate interests,” such as “plurality of the media”.  On the other hand, the Commission may consider the impact of a loss of competition on media plurality in its assessment of mergers.  Mergers and acquisitions in the media industry could reduce consumer choice, resulting in a landscape where a few dominant companies could wield considerable power over democratic processes by influencing public opinion. The Commission will consider this dynamic, alongside traditional factors like price and quality, when evaluating the implications of mergers and acquisitions in the media sector, as well as related sectors such as artificial intelligence.

Mergers can also significantly impede competition in labor markets by shifting the balance of power between employers and workers.  Monopsonies in labor markets can lead to lower wages, higher unemployment, and worse working conditions, as well as lowering downstream output and higher prices. While the existing HMG consider the potential effects of mergers on buyer power more generally, EU merger control assessments have not so far considered the effects of mergers specifically on labor markets. The question therefore arises whether the revised Merger Guidelines should provide guidance on when an expected significant loss of competition through the exercise of buyer power in labor markets leads to a significant impediment to competition.

Mergers may also raise concerns about job losses due to restructuring and offshoring. These effects are not the result of a change in market power.  Thus, they are not covered by the EUMR and will not be addressed in the revised Merger Guidelines. The In-depth Questionnaire also notes that cost savings resulting from restructuring or offshoring are unlikely to be passed on to consumers and thus should not be accepted as efficiencies.

Conclusion.  The Consultation calls for evidence on a wide range of topics.  The review process will last over two years before revised Merger Guidelines are finally adopted.  However, the Consultation provides useful background on issues the Commission considers important and some strong hints on its direction.  Changes emerging from the review process may be seen in practice well before new Merger Guidelines are finalized.

From a structural perspective, it seems likely that the current HMG and NHMG will be replaced by a single set of Merger Guidelines addressing horizontal, vertical and conglomerate mergers.  A holistic approach should lead to a more nuanced analysis of transactions that do not fit neatly into a horizontal or non-horizontal box.  The revised Merger Guidelines will also likely include more detailed treatment of competitive metrics beyond sales shares and market concentration.  The revised Merger Guidelines may also address market features such as investment cycles and innovation intensity, as well as applicable regulatory frameworks, to the assessment of notified transactions.

The Commission also apparently aims to provide more guidance on parameters raising red or yellow flags.  Such guidance would help antitrust advisors identify transactions likely to be challenged.  On the other hand, efforts to develop presumptions of anti-competitive effects based on market shares or other individual parameters will no doubt be controversial.

The revised Merger Guidelines can also be expected to provide more detailed guidance on hot-button issues such as so-called killer acquisitions, the role of ecosystems in digital and tech transactions, plurality and artificial intelligence in the media sector, privacy as a parameter of competition, the assessment of mergers in the defense sector and buyer power in labor markets.

Perhaps most importantly, the Consultation reflects the Commission’s efforts to integrate broader policy considerations set out in EVP Ribera’s Mission Letter, the Political Guidelines and the Competitiveness Compass, as well as the Draghi Report, into EU merger policy.  The Consultation notes the link between criteria such as innovation, resilience, sustainability, security and other policy goals and the EU competitive landscape.  Although many of these issues are addressed in other EU regulatory frameworks, the Commission is considering how EUMR review can be harnessed to promote EU objectives.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398

In December 2016, the Swiss Competition Commission (ComCo) fined HCl Solutions AG (HCI) around CHF 4.5 million for abuse of a dominant market position. HCI, a subsidiary of Vifor Pharma Participations AG (Vifor), operates, among other things, the “Compendium” of electronic drug information and user-specific INDEX databases (e.g. “medINDEX” for doctors), which are used via corresponding software solutions from third-party providers. In this context, ComCo accused HCI of having systematically used contractual clauses with software companies for several years that were aimed at hindering competing database providers. In addition, the inclusion of drug information in the INDEX products was only offered to pharmaceutical companies in a package (“bundled”) with additional services. In January 2022, the Federal Administrative Court confirmed the abuse of a dominant market position in principle, but reduced the sanction.

In its ruling of 23 January 2025 (2C_244/2022), the Federal Supreme Court partially upheld the appeal filed by Vifor and HCI and referred the case back to the Federal Administrative Court for a new assessment and determination of the sanction. The Federal Supreme Court’s comments on Art. 7 para. 2 CartA[1] and the sanction are particularly interesting. Here are the most important points:

No abuse of a dominant market position under Art. 7 CartA in the case of a purely hypothetical threat to competition

Although the Federal Supreme Court confirmed that HCI held a dominant position on the relevant markets, it clarified that Art. 7 para. 2 CartA is not an endangerment offense: It specified that, in accordance with the “effects-based approach“, a particular conduct must actually be potentially capable of causing harm to competition. The risk of adverse competitive effects must actually exist based on all the specific circumstances; a merely hypothetical risk of harm to competition is not sufficient. Similarly, the mere fact that a contractual clause corresponds to an element of Art. 7 para. 2 CartA is not sufficient. The Federal Supreme Court thus follows the more recent case law of the ECJ, which also follows an “effects-based approach” (see judgment of the ECJ of 19 January 2023, C-680/20, Unilever Italia).

Against this background, the Federal Supreme Court ruled as follows on the four types of conduct by HCI in question:

• Exclusive purchasing clause in a single contract (clause A): The Federal Supreme Court denied an effective capability of this clause to exclude competitors, as it only occurred in one of 176 contracts with software houses and, according to the statement of the (one) software house concerned, was of little practical significance. Moreover, it did not completely exclude third-party providers. In the opinion of the Federal Supreme Court, the clause is therefore not abusive.
• Prohibition on feeding third-party data with the same or essentially the same structure as HCI’s XML structure into software (clause B): Insofar as this clause B, which was contained in 83 of around 176 contracts, went beyond the protection permitted under copyright law and also prohibited permissible imitation, the Federal Supreme Court qualified it as abusive within the meaning of Art. 7 para. 2 lit. e CartA.
• Bundling the publication of drug information with editorial and technical quality control: In the opinion of the Federal Supreme Court, this clause is not abusive, as these are not separate goods. The quality control is part of the publication service in the “Compendium” or in the INDEX databases and is typically requested jointly by pharmaceutical companies.
• Tying the publication of drug information with (optional) free upload to AIPS: The Federal Supreme Court also ruled that this clause was not abusive, as the upload was an incidental additional service with no independent economic significance. There was neither a separate market nor an independent demand for it.

Reduction of the sanction and no consideration of intra-group sales

The Federal Supreme Court confirmed the sanction only with regard to the partially abusive clause B. However, the remaining findings of the lower court were to be set aside and the sanction reduced accordingly.

Furthermore, according to the Federal Supreme Court, the intra-group sales should not have been taken into account when determining the sanction. In contrast to a margin squeeze (BGE 146 II 217), the pharmacies and wholesalers of the Galenica Group were not involved in the abusive behavior. There was therefore no abuse of the vertical group structure.

[1] Federal Act on Cartels and other Restraints of Competition (Cartel Act, CartA) of 6 October 1995, SR 251.

More from our authors:

Competition Law of the European Union, Sixth Edition by Van Bael & Bellis € 398