The oldest cliché in Silicon Valley is that during a gold rush, the people who make the most money are the ones selling shovels.

A chart published by Bank of America Global Research over the past week seemed to prove the point. It projects the combined free cash flow (FCF) of the major hyperscalers—Amazon, Microsoft, Alphabet, Meta and Oracle—falling below zero for the first time as they pour hundreds of billions of dollars into AI infrastructure. At the same time, the combined FCF of leading semiconductor companies continues to climb sharply.

The chart quickly became a Rorschach test. AI bears saw proof that Big Tech is overbuilding. AI bulls saw evidence that chip companies are becoming the most profitable businesses ever created. Both interpretations miss the more interesting point.

The economics of AI are shifting. Today, more of the value is flowing to the companies supplying the infrastructure than to the companies financing it.

But there’s an important nuance. Nvidia doesn’t sell AI—it sells the ability to deliver AI. Semiconductor revenues are ultimately a function of hyperscaler capital expenditure, not AI adoption itself. AI usage could continue growing for years while infrastructure spending slows simply because Microsoft, Amazon or Meta decide they have already built enough capacity. Infrastructure cycles are governed by annual capital budgets; software adoption compounds over much longer periods.

For Israel, that’s the real lesson. The opportunity isn’t to predict when the infrastructure cycle peaks. It’s to build the technologies that remain valuable regardless of whether hyperscaler capex grows by 70% or 20%.

Israel’s AI paradox

Bank of America’s latest Global Research report places Israel firmly in the next tier of AI economies behind the US and China. The rankings reveal a country with exceptional intellectual capital but clear physical constraints.

Israel ranks #1 globally in AI talent concentration, with AI professionals representing 2.1% of LinkedIn members. It ranks #5 in cumulative private AI investment, with roughly $19 billion invested between 2013 and 2025, #5 in the number of AI companies, and top 10 globally in long-term AI potential. The report also highlights Israel’s entrepreneurial ecosystem and R&D intensity as key competitive advantages.

The latest AWS Engines of Growth report reaches a similar conclusion from a different angle. Based on a survey of more than 3,400 founders across 20 countries, it found that 31% of Israeli startups founded in the past five years are AI-native, the highest proportion in the study, ahead of the United States. Rather than simply adding AI features to existing software, Israeli founders are increasingly building companies where AI sits at the core of the product from day one.

These rankings reinforce something Israel has demonstrated repeatedly over the past three decades. Our advantage has never been industrial scale. It has been talent density.

A small geography brings researchers, founders, engineers, multinational R&D centres, investors and early customers into unusually close proximity. Knowledge moves quickly between universities, military technology units, startups and global technology companies. That concentration of expertise continues to produce disproportionately important companies in cybersecurity, developer tools, enterprise software and semiconductors.

The problem is that AI increasingly rewards not only intelligence, but infrastructure.

The infrastructure gap

The same Bank of America report shows where Israel falls behind. While it ranks among the world’s strongest AI ecosystems for talent and entrepreneurship, it drops to #12 in short-term AI readiness, #19 in energy readiness and #22 in data-centre costs.

Those rankings reflect structural realities rather than temporary weaknesses. Israel lacks abundant low-cost energy, large-scale industrial capacity and the natural resources required to compete in the infrastructure arms race now unfolding between hyperscalers and governments. Countries such as South Korea, Canada and the UAE enjoy advantages that are difficult to replicate through venture capital alone.

None of this argues against sovereign AI infrastructure. Israel should absolutely invest in domestic compute for defence, research and national resilience, and the government’s recently announced AI strategy is an important step in that direction. But national infrastructure policy is not the same as venture investing. Israeli startups are unlikely to outperform Microsoft or Google by building hyperscale data centres or training frontier foundation models. Their comparative advantage lies elsewhere: in the software, silicon and specialised technologies that make global AI infrastructure faster, cheaper and more efficient.

That distinction matters because it defines where Israeli venture capital can create asymmetric returns.

Build the shovels, not the mines

Israel has spent decades building capabilities that fit remarkably well with the current phase of AI.

Companies such as Mellanox, Mobileye, Habana Labs and Tower Semiconductor demonstrate that Israel can become strategically important without owning the largest manufacturing base. Mellanox became critical to modern data-centre networking before its acquisition by Nvidia. More recently, Tower Semiconductor announced a $3 billion expansion in Japan, supported by government incentives, to increase production of silicon photonics for AI and data-centre connectivity. The model is instructive: Israeli engineering, global manufacturing.

The next decade of AI will require much more than larger GPUs. It will require faster networking, optical interconnects, specialised processors, memory optimisation, advanced packaging, inference acceleration, security and power management. These are precisely the kinds of engineering problems where Israeli companies have historically excelled.

The opportunity extends beyond hardware.

As foundation models become increasingly commoditised, value shifts further up the stack into orchestration, evaluation, observability, AI security, governance and vertical applications. Enterprises will use multiple models simultaneously, choosing different models based on cost, latency, privacy and performance. They will need software that routes workloads, evaluates outputs, manages permissions, protects proprietary data and integrates autonomous agents into existing business processes.

Those software layers are far less dependent on the capex cycle than chip manufacturing. They benefit from AI adoption regardless of whether hyperscalers increase infrastructure budgets by 70% or pause spending for a year. That makes them particularly attractive for Israeli founders building long-term businesses.

A playbook for Israeli VCs

If Israel’s comparative advantage is intellectual property rather than industrial scale, venture investors should align their capital accordingly.

• Avoid competing with hyperscalers on infrastructure ownership. Capital-intensive businesses built around owning compute, operating large data centres or training general-purpose frontier models face structural disadvantages in Israel. That does not make infrastructure unimportant, but it does make it a difficult place for venture-backed startups to generate differentiated returns.
• Double down on silicon and infrastructure technology. Chip architecture, silicon photonics, networking, specialised processors, inference optimisation, memory technologies and advanced packaging all allow Israeli companies to participate directly in the global AI buildout without replicating the industrial footprint of Taiwan, South Korea or the United States.
• Own the orchestration layer. As enterprises move from experimenting with AI to operating fleets of models and autonomous agents, orchestration, evaluation, observability, AI security and governance become increasingly strategic. Israel’s exceptional concentration of AI talent makes it well positioned to build these control layers rather than another commodity application.
• Focus on sectors where technical depth matters more than scale. Cybersecurity, defence, healthcare, industrial software, financial services and robotics reward proprietary technology, domain expertise and trust. These markets are less likely to be won by whoever owns the largest compute budget and more likely to be won by whoever builds the most reliable solution.

For Israel, the answer looks increasingly clear. Our future won’t be determined by the number of GPUs installed within our borders or the size of our data centres. It will be determined by how much of the global AI stack is designed here.

Countries with abundant energy and industrial capacity will build many of the world’s AI factories. Israel doesn’t need to win that race.

It should focus on designing the chips, networking, security, orchestration software and specialised applications that make those factories work. In the AI era, the biggest opportunity may not be owning the mine—it may be designing the shovels everyone else depends on.

The original version of #FIRGUN was published in LinkedIn to 8,100+ subscribers. Sign up to get the next edition in your inbox.

Firgun (pronounced ‘feer-goon’) is an Israeli term describing genuine, unselfish joy in someone else’s success… It’s basically the opposite of “schadenfreude”. Every week since March 2020, we celebrate the wins of our startup community. Read more about why Firgun matters. If this was forwarded to you and you’d like to subscribe, previous editions are on VC Cafe.

The US-Iran ceasefire seems to have ended this week, raising questions about the implications for Israel. On the tech front, the H1 2026 reports are out and they are telling the story that’s been brewing over the past year or so – more capital concentrating around less companies. At the same time, the government is leaning further into industrial policy: from sovereign AI infrastructure and compute to new support measures aimed at keeping high-value R&D in Israel as the strong shekel increases local operating costs. Interesting nugget: Israeli tech has already produced more new unicorns in the first half of 2026 than in any full year on record.

In the wider tech world, OpenAI I broadly released its GPT-5.6 family, comprising the flagship Sol model and the cheaper Terra and Luna variants, alongside ChatGPT Work, a desktop agent that can draw context from tools including Slack, Gmail, Google Drive and CRM platforms to create documents, spreadsheets and applications. Meta launched Muse Image, the first image-generation model developed by Meta Superintelligence Labs, across Meta AI, Instagram and WhatsApp, and previewed Muse Video, as the company seeks to translate its massive AI investment into consumer products and advertising tools. SpaceX AI and Cursor were preparing to release their first jointly developed AI model, after reportedly delaying the launch to improve its efficiency, while OpenAI Chief Futurist Joshua Achiam announced that he would leave the company after nearly nine years. Anthropic continued its expansion beyond coding through Claude Cowork and announced plans to double its New York workforce to approximately 1,000 employees, taking a 16-storey building in Lower Manhattan as the frontier-model companies compete for talent and enterprise customers. Microsoft t announced 4,800 job cuts, representing roughly 2.1% of its workforce, including a major restructuring of Xbox that will eliminate thousands of gaming roles and divest as many as five studios as the company shifts more capital and management attention toward AI. Meta also plans to begin production of its internally developed Iris AI chip in September, working with Broadcom and TSMC as it attempts to reduce its reliance on NVIDIA and double its computing capacity to 14GW by 2027. The scale of the infrastructure race continues to climb, with quarterly capital expenditure by Amazon, Google, Microsoft and Meta estimated to have reached approximately $168 billion, up 74% year over year, as investors increasingly question when the unprecedented spending will translate into sustainable AI revenue and profits. In the chip market, SK hynix raised $26.5 billion through its Nasdaq debut, the largest US listing by a foreign company, after the offering was reportedly more than seven times oversubscribed, reflecting extraordinary demand for the high-bandwidth memory used in NVIDIA‘s AI systems. Micron Technology expanded its planned US investment to approximately $250 billion through 2035, covering memory manufacturing, advanced packaging and research, while Apple reportedly agreed to purchase more than $30 billion of US-made chips from Broadcom over five years. The mega-rounds also kept coming: SambaNova systems raised $1 billion at an $11 billion valuation in a Series F, strengthening its challenge to NVIDIA in enterprise AI infrastructure; Blue Origin was reported to be raising $10 billion at a $130 billion pre-money valuation in what would be the space company’s first outside financing; distributed AI research platform Primeintellect raised $130 million at a $1 billion valuation; and AI-chip startup Positron AI entered talks to raise roughly $750 million as investors continued pouring capital into alternatives to the dominant GPU providers. The concentration is becoming increasingly stark: PitchBook data indicated that US venture investment reached approximately $413 billion in the first half of 2026, with AI companies attracting the overwhelming majority of the capital and seven startups completing rounds of at least $1 billion during the second quarter alone.

QUOTE OF THE WEEK

“Imagine if Silicon Valley was a country… that’s what Israel is. An incredible density of talent, IP and IQ” – Bill Ackman

A NOTE FROM OUR PARTNERS: HUBSPOT FOR STARTUPS*

Exclusive Offer for Israeli Startups: Unlock your AI tech stack with 90% off HubSpot through the HubSpot for Startups program.

Access the HubSpot for Startups AI Tech Stack: an all-in-one bundle of exclusive GTM discounts and AI tools designed to help your startup grow faster.

Build your startup on top of industry-leading tools, including HubSpot, Claude, Clay, Notion, Lovable, Gamma, AWS, Reddit, TikTok, LinkedIn, Google, and more.

Redeem your offer here: https://offers.hubspot.com/vc-cafe-ai-tech-stack

*(sponsored)

NEW FUNDING ROUNDS

• Congratulations Ido Bar-On and team Skapion on securing a $36M seed round to deliver a a native counter drone-swarm system designed specifically for simultaneous engagements!
• Congrats Tal Shoham and team Velocity on your $27M seed round to place ads in AI applications!
• Well done Stav Levi Neumark and team Alta | AI GTM System of Actions on your $25M series A to build the AI GTM architecture for revenue teams!
• Kudos Eyal Azoulay and team Tangos on coming out of stealth with $20M in seed funding to automate financial crime investigations!
• Way to go Ben Volkow and team QIZ Security on coming out of stealth with $17M in seed funding to expand your cryptographic posture and post-quantum security readiness!

EXITS

• Congrats Izhak Zimmermann and team Attribute – FinOps Without Tagging on the acquisition by DoiT for an undisclosed sum to provide real-time, kernel-level cloud shared cost attribution that doesn’t depend on anyone reporting anything!
• Well done Tomer Shiran and team Dremio on completing the acquisition by SAP for an undisclosed sum of your open, high-performance data lakehouse platform!
• Congrats Yoely Or and team Cando Drones on going public at the Tel Aviv Stock Exchange – ?????? ??????? ??? and raising $6.3M to develop and manufacture drones for dual use!

LINKS FOR YOUR BROWSER

VC CAFE/ ISRAEL VC/ REMAGINE VENTURES

• Every founder eventually looks for the exit sign – not all exits look like Wiz. For venture-backed startups, the path usually ends in one of three ways: IPO, acquisition or shutdown. But most acquisitions are not simple “success stories” or “failures.” They are a mix of timing, strategic fit, market consolidation, talent, leverage and preparation
• The Task Economy: Why AI’s Next Big Market May Be Human Work, Not Tokens – The first wave of AI was measured in tokens. The next wave may be measured in tasks: expert judgment, workflow completion, labelled data, compliance, payouts and human-in-the-loop quality control. For Israeli founders, this is not just another freelance marketplace. It is a new AI supply chain.
• We’ve made 15 investments with Remagine Ventures II already, but we’ve yet to tell our story…. With Fund 2, our aperture is much wider. We’re also spending real time in categories we previously didn’t cover: defence tech, hardware, quantum and other frontier areas where Israeli founders have an unfair advantage.

ISRAEL

• Israel tops global rankings for AI-native entrepreneurship – Report by AWS finds nearly one in three young Israeli startups is built around artificial intelligence, outpacing the United States and other leading technology hubs.
• How Israeli start-ups are helping ships navigate a contested Hormuz – Cheap Iranian drones and GPS jamming have choked the world’s most vital oil route. A handful of Israeli start-ups think they can help ships find a way through
• How Israeli tech is powering the world cup – interestingly, it’s more defence tech than sports tech!

GLOBAL

• US venture funding reached a record $412.7 billion in H1 2026 – PitchBook and NVCA data shows AI captured 86% of all US venture investment, while more than 81% of capital went into rounds of $100 million or more. A striking illustration of how concentrated the current boom has become.
• UK Innovation Update – H1 2026 – Dealroom.co and HSBC Innovation Banking found that UK startups raised $17.1 billion in the first half, with AI companies accounting for roughly three-quarters of the total. The UK remained Europe’s largest venture market, but domestic investors participated in only a small share of the biggest rounds
• Global startup investment reached a record $510 billion in H1 2026 – Crunchbase‘s half-year review shows global startup funding already exceeded the whole of 2025, alongside a recovery in venture-backed IPOs and acquisitions

MEDIA OF THE WEEK

Israeli tech has already produced more new unicorns in the first half of 2026 than in any full year on record. According to an IVC Data and Insights analysis, 16 Israeli tech companies reached unicorn status during H1/2026, with a combined first valuation of $25.3 billion.

Israel just ranked #1 in the world for AI-native startups. According to a new AWS study of 3,400+ founders across 20 countries, 31% of Israeli startups founded in the past five years are AI-native: ahead of the US (30%), France and Japan (28%). (source)

Nice find by a #FIRGUN newsletter reader, spotted on the Cyera jobs page :-)

Bessemer Venture Partners observes five frontiers that they think will define investing in AI infrastructure :1) Harness” infrastructure: memory, context, state, eval, and observability 2) Continual learning systems: models that keep learning post-deployment 3) Reinforcement learning platforms: environments, RL-as-a-service, and RL infra 4) The inference inflection point: production-scale serving and edge deployment 5) World models: AI that understands physical and spatial reality (source)

Europe’s most valuable AI startups

Goldman Sachs maps where $7.6 trillion in AI spending is going. Goldman Sachs expects AI capital spending to reach $765 billion in 2026 and grow to $1.6 trillion annually by 2031, totalling $7.6 trillion over six years.

Narrative violation – AI adoption is accelerating hiring: A new study of more than 21,000 U.S. companies found that businesses investing most heavily in AI increased their workforce by 10% over the past two years, while entry-level hiring rose 12%. The findings suggest AI is helping companies grow faster and create new roles rather than simply replacing existing jobs. (source)

An analysis of 8,799 rounds by Carta in H1 2026 shows that valuations have gotten higher, rounds have gotten bigger, but for fewer companies.

Topic density: Guy Fighel classified 558 of the 569 sessions at the AI Engineer World’s Fair and put a cross-cutting technique lens on top of the official tracks. The biggest cluster, roughly 1 in 5 sessions, is agentic SDLC: code review, self-driving codebases, harness engineering. And the recurring theme there is: (not model quality!) verification. Code review has become the tightest constraint in the whole system, to the point that “harness engineer” is now a real job title. (source)

Finally, Pershing Square Capital founder Bill Ackman, on why Israel is a good investment…

Thanks for tuning in. There’s a lot to celebrate in our “small” community! Keep on ceating!

Eze Vidra

How many tokens are generated? How much does inference cost? How fast are token prices falling? Which model gives you the best output per dollar? The rising consumption of tokens and costs have also been the source of controversy as more US companies turn to Chinese Open Source models to save costs.

That metric still matters. But it may not be the whole story.

A recent thread by Benchmark Partner Everett Randle framed the next shift as the rise of the Task Economy: a market where the scarce input is not just compute or tokens, but the human work needed to make AI useful in the real world. Inference tokens have become the default proxy for tracking AI growth, but the next trillion-dollar category may be built around data and tasks, not tokens alone. But wait a second, wasn’t AI supposed to automate work?

Yes. But to automate work, you first need to define it, price it, evaluate it, supervise it and continuously improve it. That creates demand for a new kind of infrastructure: platforms that break knowledge work into tasks, match those tasks with experts, capture their judgment, and turn that judgment into training data, evaluations, workflows and eventually automated services.

This is where companies like Mercor, Scale AI, Surge AI, expert networks like GLG, Fiverr, Deel and vertical AI-native services start to look connected.

They are all part of the same shift: from hiring people to buying outcomes, and from buying outcomes to training machines to deliver them.

From SaaS to services to tasks

Earlier this year I wrote on VC Cafe about AI-native services: companies that do not sell software seats, but sell completed work. Not software that helps the accountant. Software that closes the books. Not software that helps the recruiter. Software that finds and qualifies the candidate. Not software that helps the broker. Software that gets the customer insured.

The task economy is the next layer underneath that thesis.

If AI-native services sell the finished work, the task economy is the supply chain that makes that work possible.

A legal AI service still needs lawyers to create examples, judge edge cases and review high-stakes outputs. A healthcare AI service still needs clinicians to label symptoms, evaluate responses and define safe escalation paths. A self-driving car company still needs humans to label road scenes, identify corner cases and review ambiguous footage. A customer support AI company still needs real human operators to define what “good resolution” looks like.

This is why Sequoia’s “services are the new software” thesis matters. Julien Bek argued that replacing an outsourced service with an AI-native services provider is much cleaner than replacing internal headcount, because the customer already accepts that the work can be done externally and already has a budget line for the outcome.

YC made a similar point in its 2026 requests for startups, highlighting AI-native service companies in areas like insurance brokerage, accounting, tax, audit, compliance and healthcare administration. YC’s point was simple: services spend is much larger than software spend, and many services are already outsourced.

The missing piece is that “services” are not delivered as one big blob. They are delivered as thousands of tasks.

That is the atomic unit.

Mercor as the clearest signal

Mercor is probably the most visible example of this new market.

The company started as an AI recruiting platform and evolved into a marketplace for expert human work used to train, evaluate and improve AI models. In October 2025, Mercor announced a $350 million Series C led by Felicis, valuing the company at $10 billion.

By June 2026, Sacra estimated that Mercor had reached $2 billion in annualized gross revenue, up from $760 million at the end of 2025. Importantly, Sacra notes that this is gross revenue, before contractor payouts, and estimates that contractors receive 60–70% of top-line revenue.

Mercor’s CEO Brendan Foody also claimed on X that the company grew from $1 million to $2 billion in ARR in 24 months, with enterprise demand becoming the fastest-growing segment. The public evidence around that specific claim is still largely company/X-led, and some observers have questioned whether “ARR” is the right frame for a marketplace-heavy model.

But even with those caveats, the signal is hard to ignore.

This is not just AI labs buying cheap data labels. It is Fortune 500 companies, AI application developers and frontier labs buying structured human expertise at scale.

Mercor has also published benchmarks showing that current AI agents still struggle with complex professional services work. In one APEX-Agents benchmark, AI agents completed less than 25% of real-world consulting, banking and legal tasks on the first try, and only 40% after multiple attempts. The harder the task, the more human planning, context and file navigation still mattered.

That is exactly why the task economy exists.

The models are getting better, but the path to improvement still runs through expert human judgment.

Why this is not Mechanical Turk

It is tempting to compare the task economy to Amazon Mechanical Turk. That would miss the point.

Mechanical Turk was built for small, generic, low-context microtasks. The new task economy is built for high-context, domain-specific, often regulated work.

The early internet data-labeling model was: “Can a crowd worker tag this image?”

The AI-native task model is: “Can a former investment banker evaluate this financial model?” “Can a lawyer identify whether this contract clause creates real risk?” “Can a physician judge whether this AI-generated triage answer is safe?” “Can a Hebrew-speaking operator with local context evaluate whether this customer support answer sounds natural?” That is a different market.

In the words Everett:

Tasks are the “unit of practice” in reinforcement learning: a model is given an initial state and an environment to act in, and its behavior is scored by a reward signal/verifier. Across many tasks, those scores are aggregated into a training signal that shifts the model’s behavior toward what scored well. Strictly speaking, “task” refers to this RL post-training substrate. But I’ll use it more loosely to stand for the unit of data-driven improvement generally, since the industry is rapidly inventing new forms that data takes in the service of making models better, and candidly because Task Economy has such a nice ring to it. I also want to distinguish this category from the dated moniker of “data labeling”, which brings to mind bounding boxes and thumbs up/down for LLM responses — the market has evolved well beyond these primitives over the last couple years into much more complex & high value tasks.

Grand View Research estimates that the global data collection and labeling market was $3.77 billion in 2024 and could reach $17.1 billion by 2030, growing at a 28.4% CAGR.

At the same time, the expert network industry reached roughly $3 billion in 2025, with Inex One estimating 12% annual growth between 2023 and 2025. Corporate customers are becoming a bigger adoption driver, making up around 45% of clients by number. Those two worlds are converging.

Expert networks historically monetized access to human judgment through calls, transcripts and surveys. AI data companies monetize expert judgment through labels, rubrics, evaluations, red-teaming and training datasets. AI-native services monetize expert judgment by embedding it into workflows that eventually become more automated. The task economy combines all three.

The platform stack of the task economy

The winners in this category will probably not look like old freelance marketplaces.

They will need to solve several layers at once:

1. Task decomposition
   Turning messy business workflows into clear units of work that can be assigned, measured and improved.
2. Expert sourcing
   Finding people with real domain knowledge, not just available labor.
3. Verification and quality control
   Knowing which expert is good, which answer is correct, and which tasks require multiple reviewers.
4. Workflow and data capture
   Turning human work into reusable training data, evaluation sets, playbooks and system prompts.
5. Compliance and payments
   Managing contractors across countries, currencies, tax regimes and employment classifications.
6. Automation over time
   This is why Deel is already positioning around AI data-labeling workforces. Deel’s own materials describe AI data labeling as a use case for global contingent workforces, with emphasis on fast scaling, local linguistic and cultural expertise, compliance, onboarding and bulk payments.

That is also why Scale AI became strategically important enough for Meta to invest $14.3 billion for a 49% stake, valuing Scale at $29 billion. Reuters reported that Scale’s Remotasks and Outlier platforms recruit and manage gig workers who manually label data for AI models.

In other words, the task economy is not just a labor marketplace. It is part of the AI supply chain.

The Israel angle

Israel has a few reasons to pay attention to the task economy. It already produced one of the defining companies of the online freelance era: Fiverr. Its Q1 2026 results show both the pressure on the old marketplace model and the opportunity ahead: active buyers declined 17.8% year over year, but annual spend per buyer rose 15.4%, services revenue grew 30%, and the company is now talking about combining expert freelancers with GenAI models and agents. That is the transition in one public company: fewer simple tasks, more complex work, and more AI-enabled delivery.

Israel also has relevant infrastructure companies. Dataloop, acquired by Dell for $120 million, built technology for managing, labeling and processing unstructured data used to train AI models. Papaya Global operates in the workforce/payments layer, helping companies manage global payroll and workers across multiple countries. Calcalist reported in January 2026 that Papaya was in advanced acquisition talks at a potential $3.5–$4.5 billion valuation. Add Israel’s strengths in cyber, data infrastructure, fintech, developer tools, healthcare and enterprise automation, and the opportunity becomes clear: not to build another Mechanical Turk, but to build trusted task infrastructure for complex verticals.

The most interesting wedges for Israeli founders may be AI data operations for regulated industries; expert-in-the-loop evaluation for Hebrew, Arabic and multilingual enterprise AI; compliance and payroll infrastructure for distributed AI taskforces; data provenance for training workflows; and vertical AI-native services that start with expert humans and automate more of the workflow over time. Israel’s edge is not cheap labor. It is domain depth, speed, technical talent and the ability to turn operational pain into software.

What changes for founders

For founders, the task economy changes the startup playbook. The old SaaS question was: “What tool can I sell?” The AI-native services question is: “What work can I take off the customer’s plate?” The task economy question is more specific: “What is the smallest repeatable unit of valuable work, who can do it best today, and how does every completed task make the system smarter?”

That last question is where the moat may come from. If all you do is route tasks to people, you are a staffing company. But if every task produces proprietary workflow data, expert feedback, evaluation sets and automation loops, you may be building something much more defensible.

Pricing will likely change too. SaaS priced by seat. Marketplaces priced by take rate. AI-native services may price by outcome. The task economy may price by completed task, verified judgment, successful evaluation, resolved ticket, closed workflow or training data value. Tokens measure model usage. Tasks measure economic work.

The risk: platform work without dignity

There is also a darker side. Platform work has always created tension between flexibility and precarity, and the AI task economy could make that worse if expert work is fragmented into disposable tasks, workers are treated as interchangeable, and platforms capture most of the upside from models trained on human judgment.

Recent reporting from The Verge and Business Insider has highlighted concerns from AI training workers around instability, pay changes and the experience of training tools that may eventually replace parts of their own work. Founders in this category should take that seriously. The best companies will not just arbitrage labor. They will build trust with both sides of the market through transparent pay, clear project expectations, portable reputation, worker protections where appropriate, strong privacy controls and a credible answer to the question: “What happens to the humans once the model improves?”

If the task economy becomes a race to the bottom, it will be fragile. If it becomes a way for experts to monetize judgment, train better systems and participate in new forms of AI-enabled work, it could be one of the most important labor-market shifts of the decade.

The next AI metric

A few years ago, we measured AI progress by model size. Then we measured it by benchmark scores. Now we measure it by tokens, inference costs and revenue run-rate. The next metric may be more practical: tasks completed.

How many useful tasks can an AI system complete? How much expert supervision does each task require? How much does each completed task improve the system? What percentage of the workflow can move from human-led, to human-reviewed, to agent-led?

That is where the task economy becomes interesting. Not because humans disappear from the loop, but because for now, humans are the loop. The companies that organize, price, verify and learn from that loop may become some of the most important infrastructure companies in AI.

For Israeli founders, this is a category worth studying closely. The opportunity is not to copy Mercor. It is to ask where Israel has domain expertise, trusted talent, complex workflows and global customers willing to pay for outcomes. The next great AI company may not sell software. It may not even sell services. It may sell the task graph behind the work.

The first gets the headlines. The second is more common. The third is far more common than anyone likes to admit.

There are exceptions, but not many. Most companies do not IPO. Most startups do not sell for billions. Many never sell at all. Only a small fraction of seed-funded companies ever reach a meaningful exit, and fewer than 1% become unicorns, while the large majority of venture-backed companies never return cash to their investors at all.

And while we tend to celebrate the big headline exits: Google’s $32B acquisition of Wiz, SpaceX’s $60B all-stock acquisition of Cursor-maker Anysphere, Facebook’s $19B acquisition of WhatsApp and Intel’s acquisition of Mobileye, most acquisitions are much less under the radar.

They don’t always come with splashy headlines, iconic founder photos or generational returns. Some are strategic. Some are defensive. Some are about talent. Some are about consolidation. Some are rescue missions dressed up as M&A.

That distinction matters, because an acquisition is rarely just a financial event. It is a timing decision, a strategic decision, an emotional decision, and often, a test of how prepared the company really is.

As I’m currently working with a number of Remagine Ventures portfolio companies on their exit strategy, I thought it’d be good to list some do’s and don’ts on startup M&A and how to get prepared. Founders should understand how buyers think, what creates strategic value, and what makes a deal more or less likely to happen.

Why startups get acquired

There are several reasons a startup gets acquired. The same deal can include more than one of them.

1. The company is running out of money

This is the least glamorous version of M&A, but it is one of the most common.

A startup raises capital, builds a team, signs customers, ships product and maybe even finds early product-market fit. But growth is not fast enough to raise the next round. The market shifts. A large customer churns. A category becomes too crowded. The burn rate is too high. Existing investors are not willing to bridge indefinitely.

At that point, the goal changes.

The company is no longer optimising for the perfect strategic buyer. It is trying to preserve as much value as possible: the team, the technology, the customer relationships, the IP, the investor capital, and the years of work that went into building the business.

These deals can still be good outcomes, especially if they land the team somewhere strong and keep the product alive. But they are very different from a proactive M&A process with multiple bidders and 12 months of runway.

The less cash a startup has, the less leverage it usually has. Acquirers know this. Founders know this. Investors know this. Time becomes a negotiating weapon.

This is why timing matters so much. Waiting until the last minute rarely improves the outcome. One takeaway for founders is: if you’re off-ramping from the venture capital path (i.e. the next round is likely not viable) start planning for M&A early.

2. The startup fills a strategic gap

This is the classic strategic acquisition.

The acquirer has distribution, capital, customers, infrastructure and brand. But it lacks a specific product, capability, technical team or market position that the startup has developed.

That gap could be in AI, cybersecurity, fintech, commerce, gaming, data infrastructure, developer tools, semiconductors, robotics, defence, healthcare or any other fast-moving category. In today’s day and age, there’s no shortage of incumbents that’d be happy to have more Agentic AI firepower for example.

Large companies are not slow because they lack smart people. They are slow because they have existing products, internal politics, complex roadmaps, legacy customers, security reviews, procurement cycles and competing priorities. A startup can focus on one problem with a level of speed and intensity that is difficult to replicate inside a large organisation.

That speed has value.

Sometimes the acquirer could build the product internally, but buying saves time. Sometimes the acquirer could partner, but ownership creates a deeper moat. Sometimes the startup has unique IP, customer traction or data. Sometimes the buyer simply cannot afford to wait.

In markets where incumbents are racing to catch up, AI is the obvious current example, the window for strategic acquisitions can be short. Buyers feel urgency, a few deals happen, and then the remaining players decide whether to buy, build or partner. That window can close faster than founders expect.

3. The acquisition is really about the team

The founders and team have built something technically impressive, but the commercial path is unclear. Or the market is real, but the company is too early. Or the product does not survive the acquisition in its original form.

In these cases, the acquirer is often buying talent, taste, urgency and domain expertise. There is also a wide range of outcomes in this kind of exit. For the buyer, the question is whether this group can accelerate an internal roadmap. For the founders, the question is whether the team will have enough autonomy, executive sponsorship and incentive alignment to do meaningful work after the transaction closes.

This is especially true in areas where hiring one person at a time is too slow: AI research, cybersecurity, infrastructure, gaming, robotics, chips, creative tools, product design or frontier consumer products. Great teams are scarce. Great teams that already know how to work together are even scarcer.

Acqui-hires can sound like consolation prizes, but that is not always fair. Many important products inside large companies started with acquired teams. The question is whether the team has enough autonomy, executive sponsorship and incentive alignment to do meaningful work after the transaction closes.

For founders, this is also where the emotional side of selling becomes real. Once you sell, the company is no longer yours. You may still lead the team, influence the roadmap and build inside the acquirer, but the ownership changes. That is not a minor detail. It is the deal.

A team acquisition can be a face-saving outcome, a strong soft landing, a modest return of capital, or the beginning of something much bigger inside the buyer. The difference usually comes down to leverage: how much cash the company has left, how unique the team is, whether there are multiple interested buyers, and whether there is enough product or IP to make the transaction more than a hiring exercise.

4. The category consolidates

Some acquisitions happen because a market is real, but the standalone path becomes harder than expected.

There may be too many vendors chasing the same budget. Customer acquisition costs rise. Buyers consolidate spend around platforms. Public market comps compress. Investors become less willing to fund “yet another” company in the category. What once looked like a venture-scale standalone opportunity starts to look like a feature, a product line, or a capability inside a larger platform.

That does not mean the startup failed. It can mean the market matured.

In many categories, the first wave is experimentation: many startups, many approaches, lots of noise. The second wave is separation: a few leaders emerge. The third wave is consolidation: the winners buy missing pieces, adjacent players merge, and incumbents acquire capabilities rather than letting them become threats. Recent data backs this up, M&A has become the dominant exit route for startups, especially as IPO windows stay narrow and selective.

For founders, consolidation can be an opportunity if they move early. It can be dangerous if they wait until every buyer has already made its move.

5. The buyer wants customers, data or distribution

Sometimes the acquisition is driven by commercial assets rather than pure product or team.

A startup may have a valuable customer base, usage data, marketplace liquidity, developer adoption, transaction volume, workflow depth or community trust. These assets can be hard to recreate from scratch.

The acquirer may want to cross-sell into those customers, plug the product into a larger go-to-market engine, or use the startup’s data advantage to strengthen its own AI, analytics or automation layer.

This can be particularly attractive when the startup has strong engagement but limited ability to monetise at scale on its own. Inside a larger platform, the same product can be sold through existing channels, bundled into enterprise contracts, or embedded in a workflow with much greater reach.

The value is not just what the startup is today. It is what the startup could become with more distribution.

The role of timing

Timing is one of the most underrated variables in M&A.

The same company can be highly attractive in one market and almost unsellable a year later. The product may not change. The team may not change. The customer love may not change. But the buyer’s priorities can change.

A new CEO comes in. A corporate development team gets reorganised. A strategic initiative gets deprioritised. A large acquirer digests two previous acquisitions and pauses M&A. Public markets fall. Antitrust scrutiny rises. The acquirer’s stock drops. A competitor gets bought. The “must-have” category becomes “wait and see.”

Founders often assume that if a buyer is interested today, they will be interested tomorrow. That is not always true.

Strong offers are rarer than they look from the outside. Corporate priorities are cyclical. Urgency can disappear. The buyer that “had to own this category” in January may decide to build internally by September.

This is why a good exit is not only about price. It is about probability of close, certainty of payment, treatment of the team, integration plan, legal terms, timing and what happens if the deal drags on.

A higher headline number that never closes is not a better outcome.

How founders can prepare before they need to sell

The best M&A processes rarely start when the founder says, “We need to sell.”

They start years earlier, through partnerships, customer conversations, ecosystem relationships, investor updates, clean governance and disciplined company-building.

Founders should build for independence. But they should also understand what makes the company valuable to others.

Know your likely acquirers

Every founder should have a living map of potential acquirers.

Not a fantasy list of “Google, Apple, Microsoft, Amazon, Meta.” A real map.

Who has a strategic gap you fill? Who is buying in your category? Who is partnering with companies like yours? Who has distribution but lacks your product? Who has the product but lacks your segment? Who is under pressure from a competitor? Who has made acquisitions before? Who has the balance sheet and the internal muscle to close?

Potential acquirers are often customers, partners, channel partners, competitors or adjacent platforms. The relationship does not have to be framed as M&A. In fact, it usually shouldn’t be. The best relationships begin around product, data, distribution, integrations, joint customers or strategic partnerships.

By the time an acquisition conversation starts, the buyer should already understand why the company matters. Founders who build those relationships early are in a better position than founders trying to cold-start a process with six months of cash left. TechCrunch’s guide to startup M&A timelines makes a similar point: sourcing offers can take months, and pre-existing relationships with acquirers are strongly preferable to a cold start.

Keep the house in order

A startup does not need to run like a public company, but it should not be a mess.

Founders should keep the basics clean:

• Financials, revenue recognition and burn
• Customer contracts and renewal data
• Cap table and option grants
• Employment agreements and contractor assignments
• IP ownership and open-source usage
• Product metrics and cohort data
• Security posture and compliance materials
• Board approvals and corporate governance
• Key vendor agreements
• Litigation, disputes or regulatory issues
• A simple, organised data room

This sounds boring. It is also where deals slow down or die.

A buyer may fall in love with the product, but legal, finance, security and HR will still run diligence. If the company cannot prove it owns its IP, cannot explain its revenue, has messy employment documentation, or has promised customers things it cannot support, the buyer will use that risk against the price or walk away.

Understand what the buyer is really buying

Founders should be brutally honest about the source of acquisition value.

Is the buyer acquiring the company for the product? The team? The customers? The revenue? The data? The IP? The brand? The geographic footprint? The regulatory licence? The technology stack? The category position?

This matters because it changes the process.

If the buyer wants the team, retention packages and founder employment terms will matter. If the buyer wants the product, roadmap and integration will matter. If the buyer wants revenue, customer quality and churn will matter. If the buyer wants IP, ownership and defensibility will matter. If the buyer wants data, consent, privacy and rights to use the data will matter.

Founders should not assume the buyer values the same things they value. The buyer’s thesis drives the diligence.

Create competitive tension where possible

The best M&A outcomes usually involve more than one credible buyer.

This does not mean running a loud auction. In fact, most startup M&A processes are highly confidential and targeted. But it does mean avoiding a single-threaded process where one buyer controls the timeline, price and terms.

If there is only one buyer, the buyer has leverage. If the company is almost out of cash, the buyer has more leverage. If the founder has already signed an LOI with exclusivity before negotiating key terms, the buyer has even more leverage.

Once a startup signs a letter of intent, leverage often shifts toward the acquirer. Legal advisers who work on the sell-side are blunt about this: signing an LOI can mean giving up nearly all of the seller’s remaining bargaining power, since the exclusivity clause is typically the only strictly binding term — the buyer can still walk away, re-trade the price, or change terms during diligence, often without penalty. The buyer gets exclusivity, runs deeper diligence and may try to renegotiate terms based on new findings. Founders should lock down the important economic and legal points before exclusivity wherever possible, and keep any exclusivity window as short as they can.

This is where investors, board members, lawyers and advisers can help. They can pressure-test offers, introduce other buyers, manage governance and push back on unreasonable terms.

Know when to bring in a banker or broker

Here’s a stronger version of that section:

Know when to bring in a banker or broker

Not every startup needs a banker.

In fact, many startups would struggle to secure a serious M&A adviser if the potential transaction is not perceived to be large enough, strategic enough or competitive enough. And getting the wrong banker can sometimes do more damage than good.

For smaller acqui-hires, distressed sales or highly specific strategic conversations, a banker may not add much value. If there is only one obvious buyer, or if the transaction is really about placing the team, the process may be better handled by the founders, investors, lawyers and board.

But above a certain size, or when there are multiple credible acquirers, a good banker can be very helpful. The right adviser can structure the process, create competitive tension, manage outreach, keep buyers moving at the same pace, help prepare materials and free the founders to keep running the business.

A banker does not magically create an exit. If there is no strategic value, no buyer urgency, no credible business and no real acquirer universe, a banker cannot manufacture a great outcome.

But in the right situation, an adviser can help founders avoid common mistakes: talking to the wrong people, sharing too much too early, letting one buyer dictate the pace, failing to prepare the data room, agreeing to exclusivity too soon, or underestimating how long the process will take.

The important part is choosing carefully.

A good banker understands the category, has real buyer relationships, knows how strategic acquirers think and is willing to be honest about what is achievable. A weak banker may overpromise, spam irrelevant buyers, create noise in the market, waste management time or make the company look less attractive than it is.

Founders should also pay close attention to the engagement letter.

One common issue is the tail. Bankers often include a clause that entitles them to a success fee if the company is acquired within a certain period after the engagement ends, especially if the buyer was introduced or contacted during the process. A tail is standard, but the details matter: how long it lasts, which buyers it covers, whether the buyer list is clearly defined, and what happens if the founder already had a prior relationship with that acquirer.

Other terms matter too: monthly retainers, minimum fees, exclusivity, reimbursement of expenses, termination rights and what counts as a successful transaction. These details can become painful later if the process changes or if a buyer emerges through a different channel.

M&A is not just a transaction. It is a process. And the wrong process, run by the wrong adviser, can reduce leverage rather than increase it.

The best banker is not the one who promises the highest price. It is the one who can help create a credible, disciplined process while keeping the company’s options open.

Manage cash like the process may take longer than expected

Founders should assume an acquisition process will take months, not weeks.

Even a motivated buyer needs internal approvals, product diligence, financial diligence, legal diligence, security review, HR review and executive sponsorship. If the buyer is public, regulated, international or acquisitive, the process can become even more complex.

A useful rule of thumb is to plan for enough runway to run a real process and still survive if the deal falls apart. Advisers who run these processes regularly suggest planning for up to nine months from start to close, including sourcing offers, evaluating bids, diligence, signing and closingm even when a motivated, tightly-run process could get through diligence in a few weeks.

That may sound conservative, but it reflects reality: time kills deals. The longer a process drags on, the more room there is for market changes, internal politics, diligence findings, re-trading or buyer fatigue.

If the company has only a few weeks of cash left, the process is no longer strategic. It is distressed.

Build for independence, understand your value to others

The best acquisition strategy is to build a company that does not need to be acquired.

Strong products, happy customers, clean metrics, low burn, credible growth and a motivated team all create optionality. Optionality is leverage.

But founders should not be naive. Venture-backed startups operate in markets shaped by capital cycles, buyer urgency, investor expectations and strategic windows that open and close. An exit is not always available when you want it. A buyer is not always there when you need one. The perfect offer may never come.

The best founders build for independence. But they also understand what makes their company valuable to others: the product, the team, the customers, the IP, the data, the category position, or the strategic gap they fill for a much larger player.

In venture, the exit sign eventually appears.

The founders who are prepared are usually the ones with the best chance of choosing which door to walk through.

The original version of this newsletter was published on LinkedIn. Subscribe to join 8,100+ people who got their copy via email today.

Firgun (pronounced ‘feer-goon’) is an Israeli term describing genuine, unselfish joy in someone else’s success… It’s basically the opposite of “schadenfreude”. Every week since March 2020, we celebrate the wins of our startup community. Read more about why Firgun matters. If this was forwarded to you and you’d like to subscribe, previous editions are on VC Cafe.

Yesterday (July 2nd) marked 1,000 days since October 7th 2023. As we celebrate another week of innovation and entrepreneurship, it’s important to pause and remember what Israel has carried through these 1,000 days, and how much strength it takes to keep building.

According to the preliminary findings by IVC Data and Insights and LeumiTech, Israeli tech companies raised $7.6 billion in 193 rounds in H1/2026, up 52% from H1/2025. Q2 alone accounted for $4.2 billion in 86 rounds. AI infrastructure, cybersecurity, healthcare and defense tech leading the way. CTech by Calcalist recorders a slightly higher number – Israeli startups raised about $8.4 billion spread across a lower number of disclosed deals.

As America celebrates the 4th of July, it’s a good moment to remember that beyond the headlines, the U.S.–Israel relationship remains one of the most important strategic partnerships in the world. It is built on shared democratic values, deep security cooperation and a relentless belief in innovation, resilience and freedom. The news cycle may be noisy, but the fundamentals of the relationship remain strong.

In the wider tech world, Global venture funding reached a record $510 billion in H1 2026, with AI accounting for an unprecedented share of capital as investors continue concentrating behind category leaders. Baseten announced a $1.5 billion funding round at a reported $13 billion valuation, reinforcing that inference infrastructure has become one of the hottest segments in AI. Groq raised another $650 million to expand its AI inference cloud. The hyperscalers continued to double down on AI: Microsoft said it will invest $2.5 billion into Microsoft Frontier Co., a new operating business that is designed to embed 6,000 “AI and engineering experts” inside of customer organizations. ; Amazon said Project Kuiper / Amazon Leo now has nearly 400 satellites in orbit ahead of commercial service; Anthropic‘s Fable 5/ Mythos 5 is back to non-US customers. The company is reportedly working with Samsung Electronics on custom AI chips; and OpenAI made headlines with reports that it has discussed giving the U.S. government a 5% equity stake as part of a broader AI competitiveness initiative. On the regulatory front, Google lost its final appeal against the EU’s Android antitrust ruling, leaving a €4.1 billion fine in place for using Android to block rivals. X introduced Live Studio, a new livestreaming command center.

QUOTE OF THE WEEK

“Far and away the best prize that life offers is the chance to work hard at work worth doing.” — Theodore Roosevelt

A NOTE FROM OUR PARTNERS: HUBSPOT FOR STARTUPS*

Exclusive Offer for Israeli Startups: Unlock your AI tech stack with 90% off HubSpot through the HubSpot for Startups program.

Access the HubSpot for Startups AI Tech Stack: an all-in-one bundle of exclusive GTM discounts and AI tools designed to help your startup grow faster.

Build your startup on top of industry-leading tools, including HubSpot, Claude, Clay, Notion, Lovable, Gamma, AWS, Reddit, TikTok, LinkedIn, Google, and more.

Redeem your offer here: https://offers.hubspot.com/vc-cafe-ai-tech-stack

*(sponsored)

NEW FUNDING ROUNDS

• Nice milestone Guy Shimoni and team Shortical on securing $100M in user acquisition financing to scale your short form drama app reach!
• Congratulations Gal Aga and team Aligned on your $60 million Series B for your AI B2B sales workspace to help manage complex sales processes from first call to close.
• Well done Yadin Soffer and team Traysar Industries on emerging from stealth with $25 million in seed funding to develop autonomous systems for underground military environments.
• Way to go Erez Riahi and team Esh-Tech Systems on securing an $18M round to accelerate the deployment of DroneLight, an anti-drone laser defence system!
• Great stuff Shahar Erez and team arato.ai on coming out of stealth with a $10 million seed round to build an AI testing platform that simulates user interactions to identify risks in AI-driven products.
• Kudos Guy Eisenkot and team Baz on your $9 million seed extension (bringing total funding to $17 million) to scale your tools to review AI-generated code before bugs and security flaws reach production.
• Mazel tov Kim Aviv and team Mila on securing $2.5M pre-seed funding for your design-led intimate wellness brand!

EXITS

• Congratulations Sivan Rauscher Ganot and team SAM Seamless Network on the acquisition by Qualcomm for more than $150 million to protect over 500 million connected devices across 15 million networks!
• Well done Maor Ezer and team ai.work on the acquisition by ServiceNow for tens of millions of dollars for your AI agent platform for internal enterprise service and operational workflows.
• Way to go Ian Riopel and team Root Security on the acquisition by Belgian cyber unicorn Aikido Security for an estimated $70–100 million. Aikido will open a development center in Israel following the deal.
• Kudos Eran Shir and team Nexar Inc. on the merger with U.S.-based Nauto to create a global physical AI company focused on real-world driving systems. The combined platform will use billions of miles of driving data to train autonomous and fleet intelligence models.
• Not all exits have a happy ending, but they are nevertheless an outcome. This is the case for Itamar Jobani and team PayEm which was acquired this week by Top Group Holding for an estimated $4M in total for their expense management platform.

NEW FUNDS

• Kudos Yotam Alroy and team Pitlane Capital on launching a new $25M fund to buy out aging family businesses. The new fund will invest in search funds targeting small family businesses without successors.

LINKS FOR YOUR BROWSER

VC CAFE/ REMAGINE VENTURES/ ISRAEL VC

• Israel’s $30B AI Sovereignty Bet – The government wants 100,000 GPUs, sovereign AI infrastructure and national strength in Cyber AI, Physical AI and deepfake defense. The opportunity for startups is in the layers Israel can actually win.
• Diligence Before Data – What pre-seed investors look for when there are no metrics yet
• What Israeli Founders Need to Know About Raising in H2 2026 – The honest state of early-stage funding: Data-driven signals from 128 tracked rounds in H1 2026
• What Israel Can Learn From Asia’s Innovation Playbook – Japan, Korea, Taiwan and Singapore show that the next phase of innovation is not just about producing startups. It is about turning startups, talent, infrastructure and national strategy into durable technological advantage.

ISRAEL

• Israel’s tech sector raises $8.4 billion in the first half of 2026 – by CTech by Calcalist 129 funding rounds spanning AI infrastructure, cybersecurity, healthcare and defence signal sustained investor appetite despite an ongoing war and global economic uncertainty
• Israel attracted 9% of the global investment in Quantum technology – on The Times of Israel
• Wix saw it’s acquisition Base44 cross the $100M in ARR within one year. But now, the company developed its own coding models and it’s changing everything.

GLOBAL

• How an AI Bust Could Ripple Through The Global Economy – Plus, South Korea’s chip giants plot new plants, and AI brings bumper profits for memory producers
• McKinsey Quantum Technology Monitor 2026: A commercial tipping point by McKinsey & Company
• Are Mega-Funds Taking Over Seed? Murph Capital analyzed 20+ mega-funds transforming the early-stage landscape and unpacked how their AI-era allocations, lead rates, and pricing power really work

MEDIA OF THE WEEK

The first official H1 2026 fundraising snapshot by IVC Online and Leumitech – Israeli tech companies raised $7.6 billion in 193 rounds in H1/2026, up 52% from H1/2025. Q2 alone accounted for $4.2 billion in 86 rounds. Cyber companies led fundraising with $2.57 billion. Defense Tech and Quantum companies raised $846 million in just six months, approaching the $953 million raised during all of 2025. (source)

Israeli tech just posted its strongest 18 months in years and the data backs it up, not just the headlines by Israel VC (source) ]

Crunchbase News reports that global venture funding hit a record $510 billion in H1 – with OpenAI and Anthropic alone accounting for 43% of the total – as IPOs and M&A rebounded and billion-dollar rounds spread into AI infrastructure, defense, robotics, and healthcare. (source)

Bessemer Venture Partners’ Defense Tech Roadmap (source)

The top 20 startups by web traffic since 2020 by Menlo Ventures Deedy Das (source)

How professional services firms are building their AI strategies by CB Insights

The Loop infrastructure market map (source)

The industrials and manufacturing B2B software innovators landscape by Insight Partners

AI is creating an explosion in solopreneurs, according to a new report by Stripe (source)

Watch Palantir Technologies CEO Alex Karp talk about Israel this week on CNBC

Thanks for tuning it. That’s all for this week. There’s a lot to celebrate in our “small” community. Keep on creating!

Shabbat shalom and happy 4th!

Eze Vidra

But Israel is not the only small or medium-sized economy trying to turn innovation into national advantage. Across Asia, countries with very different political systems, industrial structures and histories have been asking a similar question: how do we build technological capabilities that matter not just for startups, but for national resilience, economic competitiveness and geopolitical relevance?

This is the core idea I took from Ramon Pacheco Pardo and Robyn Klingler-Vidra’s Foreign Affairs essay, “The Secret to Japanese and South Korean Innovation,” and from their broader argument in Startup Capitalism. Full disclosure: Robyn is my wife, so I’m not exactly a neutral reader. But the book’s core idea is especially relevant for Israel: startups in East Asia are not always framed as rebels that must disrupt incumbents. They are often seen as resources for large companies, tools of industrial renewal and instruments of national strategy.

That is a very different mental model from the classic Silicon Valley mythology: a brilliant founder, a garage, a VC cheque and a mission to destroy the old order. Israel adopted a version of that model extremely well. It helped create a thriving venture ecosystem. It also produced some unintended consequences: too many companies built for early acquisition, too much dependence on foreign capital, not enough local scale-up capacity, and a gap between startup creation and national industrial depth.

Asia offers a useful counterweight.

The common thread: innovation as national capability

Japan, South Korea, Taiwan, Singapore and Israel are very different places, but they share several features that make the comparison useful. They operate in competitive and sometimes threatening geopolitical environments. They cannot rely on natural resources alone. They depend heavily on exports, talent and global markets. They understand that technology leadership is not only about valuation, but about sovereignty.

This is increasingly true for Israel as well. The government’s new National AI Program, approved in June 2026, includes an ambitious target of 100,000 processing units, alongside sovereign compute infrastructure, a national quantum computer, human capital development and international partnerships.

That is the right direction. But infrastructure alone is not a strategy. GPUs are not an ecosystem. The harder question is how Israel connects startups, universities, government, defense, corporates and global markets into a more coherent national innovation system.

This is where Asia becomes interesting.

Japan: make large companies part of the startup system

Japan is often caricatured as slow, corporate and risk-averse. But that misses what is changing. Japan has been putting startups at the centre of a much broader national growth agenda. Its Startup Development Five-Year Plan set ambitious targets: increasing startup investment to ¥10 trillion, creating 100 unicorns and supporting 100,000 startups, while also promoting open innovation with large established companies.

More recently, Japan has gone even bigger. Prime Minister Sanae Takaichi’s 2026 economic blueprint aims to mobilise ¥370 trillion, roughly $2.3 trillion, in combined public and private investment by 2040 across strategic sectors including AI, semiconductors and space.

The lesson for Israel is not that government should pick winners from Jerusalem. The lesson is that large companies should not be treated only as acquirers of Israeli startups or foreign employers of Israeli talent. They can become design partners, distribution channels, manufacturing partners, data partners and anchor customers.

Israel has world-class founders but relatively few large domestic technology companies. That makes the multinational R&D presence important, but also incomplete. Israel needs stronger mechanisms to connect startups with local and global incumbents around areas where the country already has depth: cybersecurity, defense tech, AI infrastructure, health, fintech, agritech, mobility, robotics, semiconductors and creative technologies.

Japan’s model suggests that open innovation works best when it is not left to networking events and corporate theatre. It requires incentives, procurement pathways, legal frameworks and senior-level commitment from industry.

South Korea: let private investors lead, but use public capital to scale

South Korea is one of the world’s most impressive innovation economies. In the 2025 Global Innovation Index, Korea climbed to 4th globally and led the world in the human capital and research pillar, as well as in business-performed R&D and researchers in businesses.

Korea’s startup strategy is particularly relevant for Israel because it does not replace private capital with government capital. Instead, programmes like TIPS use private investors and accelerators as the first filter. TIPS is designed to identify promising startups through accredited private operators, then add government R&D support, mentoring, incubation and global market assistance.

The scale is meaningful. Korea’s Ministry of SMEs and Startups says TIPS has supported approximately 5,000 startups, mobilised KRW 21.3 trillion in private investment, and contributed to 48 public listings since launch. From 2026, Korea plans to expand TIPS into a more integrated framework covering formation, scale-up and global expansion.

Korea is also becoming more explicit about connecting startups with large corporations. Its 2025 plan includes a Deep Tech Value-Up Program to match startups with the needs of large corporations, plus a Value-Up Fund that provides matching investment when large companies invest.

For Israel, this is a powerful idea. The Israel Innovation Authority has long played an important role, and the renewed Yozma Fund is already trying to bring more local institutional capital into Israeli venture. Reuters reported in January 2026 that about $450 million had been directed into local VC funds through public-private programmes, including Yozma, to support Israeli tech innovation.

But the next phase may require more aggressive public-private matching mechanisms in strategic fields. Not generic grants. Not endless committees. But structured programmes where private investors, corporates and government share risk in fields that are too capital-intensive or strategically important for venture capital alone.

AI infrastructure, chips, robotics, quantum, space, defense and bio-convergence all fall into this category. These companies need more than a seed round and a clever deck. They need labs, pilots, compute, regulatory access, procurement, patient capital and global partnerships.

Korea shows that public capital can be catalytic when it follows market signals rather than replacing them.

Taiwan: build a niche superpower, then sustain the ecosystem around it

Taiwan’s innovation lesson is not simply “semiconductors matter.” Everyone now knows that. The deeper lesson is that a country can turn a narrow technological niche into a source of global leverage if it builds the full ecosystem around it.

In a recent piece for The Conversation, Robyn Klingler-Vidra describes Taiwan as a “niche superpower”: a relatively small economy that commands outsized influence because it dominates a strategically indispensable industry. TSMC alone produces more than 90% of the world’s most advanced semiconductor chips, which are essential for smartphones, AI, high-performance computing and advanced military systems.

But Taiwan’s position was not created by TSMC alone. It was built over decades through a dense industrial ecosystem: engineering talent, supplier networks, research institutions, science parks, specialist manufacturing know-how and a deep division of labour across the chip value chain. Hsinchu Science Park became the physical cluster where universities, suppliers, chip designers, foundries and engineers could compound knowledge over time.

That is the part that is hardest to copy. Other countries can subsidise fabs. They can announce chip strategies. They can invest billions. But Taiwan’s edge comes from the accumulated ecosystem around the fabs: the suppliers who know how to respond quickly, the engineers trained in the details of yield improvement, the customers who trust the process, and the culture of operational excellence required to manufacture at the frontier of physics.

For Israel, the lesson is focus and depth. Israel does not need to own every layer of every technology stack. It needs to identify the domains where it can become a niche superpower of its own. Cybersecurity is the obvious example, but the same logic could apply to AI security, defense tech, autonomous systems, robotics, synthetic media, AI infrastructure, quantum software, computational biology or secure cloud infrastructure.

The question is not only whether Israel can produce great startups in these fields. It is whether Israel can build the surrounding ecosystem that allows those startups to compound into national advantage: talent pipelines, suppliers, local customers, procurement, testbeds, research institutions, patient capital and global distribution.

Singapore: orchestrate the ecosystem like a national platform

Singapore may be the most relevant Asian comparator for Israel because it is not a large domestic market, nor does it have the industrial depth of Japan, Korea or Taiwan. Like Israel, it has had to think globally from day one. But Singapore’s innovation model is more coordinated, more deliberately institutional and more tightly connected to national economic planning.

The best example is Singapore’s Research, Innovation and Enterprise strategy. Its RIE2030 plan commits S$37 billion over five years to research, innovation and enterprise, with emphasis on competitiveness, resilience and strategic national priorities.

Singapore also created Startup SG as a unified platform for founders, investors, accelerators and ecosystem builders. Enterprise Singapore describes Startup SG as providing funding and non-financial support to local and global startup enablers, including accelerators that help startups access markets, talent and financing.

The lesson for Israel is coordination. Israel has extraordinary startup density, elite technical talent and global entrepreneurial instincts. But the ecosystem often works despite fragmentation rather than because of orchestration. Singapore shows what a small country can do when innovation policy, talent policy, capital formation, corporate partnerships and national infrastructure are designed to reinforce each other.

As a side note, the public sector is very well regarded in Singapore. It pays well and attracts top talent from the leading universities. Part of their success in their national support plans for startups stems directly from the quality of the talent involved.

The Asian Innovation Strategy Matrix

What should Israel do differently?

Israel does not need to become Japan, Korea, Taiwan or Singapore. Its strengths are different. Israeli founders are impatient, global from day one and comfortable with ambiguity. That should be preserved. The danger is not too much entrepreneurship. The danger is assuming that entrepreneurship alone is enough.

There are five practical lessons Israel can take from Asia.

First, make corporates part of the model. Israel needs more structured collaboration between startups and large companies, both domestic and multinational. This should include procurement pilots, shared testbeds, co-development frameworks and incentives for corporates to invest in and buy from Israeli startups.

Second, use public capital to crowd in private capital. Korea’s TIPS model is a reminder that government funding works best when it leverages private selection. Israel can expand this approach in strategic deep-tech sectors where the market signal exists but the capital requirements exceed normal seed and Series A appetite.

Third, focus national strategy around areas of real advantage. Taiwan’s semiconductor strategy works because it compounds an existing strength. Israel should be equally disciplined. It cannot own the entire AI stack, but it can lead in areas where its existing advantages matter: cyber AI, physical AI, defense-adjacent autonomy, secure infrastructure, AI safety, developer tools, synthetic media, health AI and applied robotics.

Fourth, build infrastructure that startups can actually use. Sovereign compute is important, but the test is access. If only large companies and universities benefit, the impact will be limited. The best Israeli founders should be able to access compute, data, labs, sandboxes and pilot customers without spending a year navigating bureaucracy.

Fifth, update the national story. “Startup Nation” was the right brand for the 2000s and 2010s. The next phase should be about strategic innovation: companies that solve hard problems, strengthen national resilience and compete globally in markets that matter.

From Startup Nation to strategic innovation nation

The countries that will win the next decade of innovation will not be the ones with the most accelerators or the flashiest demo days. They will be the ones that can connect talent, capital, infrastructure, regulation, large customers and global markets around strategic technological capabilities.

Israel already has many of the ingredients. It has elite technical talent, strong venture capital, deep security expertise, entrepreneurial culture and global networks. But Asia shows that the next step is orchestration.

Japan teaches the importance of linking startups with large companies and national industrial priorities. Korea shows how private investors and public capital can work together. Taiwan shows the value of compounding a national strength into adjacent technologies. Singapore shows how a small state can coordinate capital, talent, infrastructure and policy as one national platform.

For Israel, the opportunity is not to abandon the Startup Nation model. It is to evolve it.

The next Israeli success story should not only be a startup that gets acquired by a foreign giant. It should also be a company that becomes part of the country’s strategic infrastructure, creates enduring local capability and sells mission-critical technology to the world.

That is the real lesson from Asia: startups are not only engines of disruption. In the right ecosystem, they are instruments of national power.