Executive takeaways

Most conversations about retail technology start with the software: the algorithm, the platform, the automation. Fewer start with the harder problem underneath it all, which is getting business teams, product teams, engineering teams, and program teams in different countries to agree on what “done” looks like. Across nearly two decades of cross-border program management, running technical programs across India, China, Belgium, and now the United States, I’ve come to believe that coordination, not code, is where most cross-border technology programs actually succeed or fail.

A career built across borders, not just across systems

My path into this work didn’t start in retail. I trained in production engineering at the National Institute of Technology, Trichy, in India, then spent close to a decade with a global IT services firm on enterprise technology programs across China and Belgium, leading SAP rollouts, logistics implementations, and multi-country Agile transformations. Each engagement added a market with its own regulatory quirks, language barriers, and operational habits that no amount of software design could paper over.

One of the clearest examples came during a SAP rollout spanning mainland China, Hong Kong, and Taiwan. On paper, it was a single template. In practice, it meant reconciling three different script conventions for the same language family: Simplified Chinese on the mainland, Traditional Chinese in Taiwan, and Traditional Chinese with Cantonese-specific usage in Hong Kong. Each market also had its own regulatory requirements, including local-language views that users in each location needed to do their jobs. A system that technically worked everywhere was still unusable if it couldn’t speak to each market in its own terms.

The same lesson showed up again on a logistics rollout in Belgium for a global consumer electronics brand, where I coordinated bonded and non-bonded warehousing, inbound and outbound logistics, and purchasing. Before the rollout, each country had run these functions differently. As I recall it, the team grew by roughly half over the course of the release. It was one of five Scrum teams implementing the same system across Europe, and it grew because the program proved it could deliver before it was handed more scope. Back in Shanghai, leading a technical program for a multinational consumer goods company, the same pattern played out at a larger scale. After we identified gaps in the client’s operations and proposed a solution to close them, the account grew by close to a third. Trust was earned the same way twice, on two different continents.

Today, I lead automation and optimization platforms for retail business and space planning across multiple international markets, working with a large cross-functional team spanning product, engineering, and program roles. The technology stack changes. The underlying challenge does not: keeping distributed teams across business, product, engineering, data science, infrastructure, and security working from the same plan, in the same language, toward the same outcome.

Why the software is rarely the hard part

A system that works cleanly in one market and breaks in the next isn’t really a working system. It’s a workaround with a good demo. Building something that holds up across multiple countries means designing for real differences in supply chains, store formats, and shopper behavior from the start, not retrofitting them in after a deadline-driven rollout.

This is also why established delivery frameworks, SAFe and ITIL among them, devote so much structure to the handoff between business intent and technical execution rather than to the technology itself. In my experience, the failure point is almost never the technology. It’s the handoff: the place where a decision made by a business team has to become something engineering, data, infrastructure, and security teams can build, test, and operate reliably, across time zones, without losing the original intent. Technical program management, done well, is the discipline of protecting that handoff.

What this looks like in practice

Over the years, I’ve relied on three practices to keep multi-market programs resilient and aligned:

• Standardized “Definition of Ready” and “Definition of Done” gateways: Teams don’t pull requirements into a sprint based on a vague commercial request. Instead, features are held to maturity models based on INVEST criteria. A feature doesn’t enter development until backend, API, and local business requirements are fully aligned, which minimizes translation loss between business intent and technical architecture.
• Synchronized cross-geographic cadences: Running cross-functional teams across time zones requires predictable, multi-tiered communication. Structured “Scrum of Scrums” sessions bring together core engineering, data, infrastructure, and security leads with regional execution teams to resolve inter-team dependencies and bottlenecks in real time.
• Automated testing and data-quality guardrails: Scaling a platform across multiple countries means local exceptions are inevitable. Instead of patching in custom overrides that pile up as technical debt, automated SIT/UAT testing pipelines verify before deployment that core platform updates don't break regional compliance or local store-format logic.

The less glamorous kind of technology leadership

Done well, this kind of program management is invisible by design. Shoppers never see the coordination behind a store’s shelf layout or the systems deciding what stocks where. They only notice, or don’t notice, whether the product they came in for is actually there. Making that hold up across multiple countries and a wide range of store formats is a less glamorous kind of technology leadership than terms like “AI-driven transformation” usually suggest. I’d argue it’s also the more durable kind, the kind that keeps working after the initial launch excitement fades.

About the author

Gyanesh Kumar Jha is a technology operations and program leader based in the United States. With nearly two decades of global experience spanning India, China, Belgium, and the United States, he specializes in technical program management, enterprise Agile transformations, and automated platform delivery across multi-country retail and CPG ecosystems. He holds a Bachelor of Engineering (Production Engineering) from NIT Trichy and an MBA from Liverpool John Moores University.

Today is Thursday, Nov. 12.

1. Nearly one-quarter of traded goods face concentrated supply risks

McKinsey Global Institute found 23% of internationally traded products face two vulnerability factors, up from 15% in 2000, reinforcing diversification as a priority amid geopolitical tension and transport disruption.

The Wall Street Journal  |  Read the original article

2. U.S. container imports set a monthly record

U.S. ports processed 2.67 million import TEUs in October, 16% more than a year earlier, as shipments from China rose 22.9% and reached their highest level since January 2025.

Reuters  |  Read the original article

3. Mexico tightens steel traceability as U.S. tariff talks advance

Mexico removed roughly 1,100 mills from its approved registry and says third-country steel and aluminum imports fell about 30%, aiming to prevent tariff evasion while seeking relief from U.S. duties.

Reuters  |  Read the original article

4. EU and U.K. weigh delaying electric-vehicle tariffs

Britain and the EU are expected to negotiate delaying 10% electric-vehicle tariffs as automakers struggle to meet post-Brexit content thresholds intended to encourage a regional battery supply chain.

Financial Times  |  Read the original article

5. BMW adds regional suppliers for next-generation batteries

BMW contracted with LG Energy Solution for North American battery cells and Samsung SDI for European supply beginning in the late 2020s, supporting localized sourcing for its sixth-generation electric vehicles.

Supply Chain Dive  |  Read the original article

6. Logistics layoffs and bankruptcies expose continuing freight stress

At least 10 transportation and logistics employers announced 1,021 job cuts, while 15 freight-related businesses filed for bankruptcy protection between mid-September and Nov. 10, according to a FreightWaves compilation.

FreightWaves  |  Read the original article

7. U.S. rail intermodal traffic reaches a monthly record

October intermodal volume rose 9.6% year over year to more than 298,000 containers and trailers per week, while total carloads increased 3.4%, according to Association of American Railroads data.

FreightWaves  |  Read the original article

8. USPS narrows its loss as Ground Advantage gains volume

USPS reduced its fiscal 2026 net loss to $6.5 billion as revenue reached $82.9 billion, while Ground Advantage volume grew 10.5% despite a 5.3% decline in total package volume.

Supply Chain 24/7  |  Read the original article

9. DHL pilots adaptive robots and live warehouse orchestration

DHL is testing robots that adjust when warehouse conditions change and deploying AI software to coordinate labor and equipment, although the company has not disclosed pilot tasks or performance results.

Supply Chain 24/7  |  Read the original article

10. Shared charging pilot tests a new model for electric trucks

Kuehne+Nagel electric trucks used Amazon chargers during scheduled U.K. stops, completing 90% of planned sessions and adding about 40% battery charge without requiring separate charging detours.

Supply Chain 24/7  |  Read the original article

Executive takeaways

Most organizations do not suffer from a shortage of supply chain metrics. They suffer from having too many. There is an abundance of dashboards, KPIs and analytics platforms available to supply chain managers today, but measures sometimes conflict, and despite the familiar belief that what gets measured gets managed, decision-making is often poorly connected to the information being generated. The challenge is not simply deciding what to measure, but designing a coherent performance measurement system.

Practice and academic research have produced hundreds of performance indicators, frameworks and methodologies. Supply chain managers can use many different approaches such as SCOR, the Balanced Scorecard, artificial intelligence, predictive analytics and digital dashboards. Yet many of these approaches focus on particular aspects of performance measurement rather than the design of the overall system. While they provide valuable guidance, they do not necessarily help organizations determine whether their performance measurement system has been designed coherently. The challenge facing supply chain managers is therefore designing a performance measurement system that supports organizational and supply chain objectives.

Analysis of more than 200 supply chain performance measurement studies revealed a simple but powerful insight: effective performance measurement is about much more than selecting metrics and dashboards. A structured analysis of these research studies identified six recurring dimensions that together form the 6Ws Framework, a practical structure for designing, reviewing and improving performance measurement systems. In practice, many organizations begin with dashboards and KPI selection. The 6Ws Framework instead begins by asking six questions about the design of the measurement system itself.

• Why are we measuring?
• What should we measure?
• Where should measurement take place?
• When should measurement occur?
• Who should be involved?
• Which tools should be used?

Collectively, these dimensions provide a common language for thinking about supply chain performance measurement and encourage managers to consider the design of the measurement system as a whole, rather than focusing exclusively on individual KPIs. They also help address a common challenge: supply chain measures, tools and frameworks are often developed and applied in isolation from one another.

The 6Ws in practice: Addressing supply chain stockouts

The value of the framework can be illustrated using a common supply chain challenge: recurring stockouts during promotional activity. Traditional responses would often focus on increasing inventory and/or adding more KPIs to existing dashboards. However, the 6Ws Framework suggests that organizations should first step back and address these six interconnected questions.

Why are we measuring?

The starting point is to clarify the objectives of measurement, as many systems begin by selecting and introducing metrics rather than defining success. Research on supply chain performance measurement shows that meaningful measurement is difficult unless its purposes are clear at an early stage. Performance measurement should identify gaps between actual and desired outcomes, understand their causes and support improvement actions. In the stockout example, the goals may be to reduce lost sales, improve product availability and protect profit margins that are affected by the stockouts, and these should be the ongoing focus of the system, rather than simply generating additional data.

What should we measure?

Once the objectives are clear, organizations should identify the indicators that best reflect performance. In the stockout scenario, measures such as forecast accuracy, case-fill rates and on-shelf availability may be useful. Effective supply chain performance measurement requires a balanced set of hard and soft indicators that collectively explain why stockouts occur and what can be done to prevent them.

Where should measurement take place?

Many organizations understandably focus on measuring within specific departments, but supply chain performance is generated collectively across multiple internal and external actors. A stockout experienced by a customer may be caused by poor supplier performance, production scheduling issues, inventory imbalances or weak retailer forecasting. In this example, measurement should therefore extend across demand planning, production, distribution centers, suppliers and possibly even customers. Looking beyond the functions and departments that are traditionally measured, as well as organizational boundaries, improves visibility and helps managers identify where problems originate rather than simply where they become visible.

When should measurement occur?

Effective performance measurement requires consideration of not only what is measured, but also of the stage of the supply chain process at which measurement takes place. Many organizations rely on historical reporting, discovering problems only after they have already affected performance. In the stockout example, measurement should take place during forecasting, production planning, promotional execution and post-event reviews. Measuring at different stages of the supply chain cycle supports the early identification of emerging issues and the implementation of corrective actions before stock becomes unavailable.

Who should be involved?

Technology alone cannot solve supply chain problems, and the research highlights the importance of individual and organizational capabilities, such as leadership, decision-making, communication, relationship management and analytical skills. In the stockout example, responsibility should not sit exclusively with inventory planners, as demand planners, supplier managers, sales and operations planning teams, logistics specialists and senior managers all contribute to performance outcomes. These roles require different competencies, and so effective performance measurement depends on ensuring that the right people have the right skills, authority and collaborative relationships necessary to interpret information and act decisively on it.

Which tools and approaches should be used?

Finally, organizations must determine which tools, technologies and frameworks will support effective performance measurement. In the stockout scenario, these might include forecasting software, ERP systems, SCOR-based process measures, digital dashboards or AI-enabled planning systems. However, organizations often focus too heavily on tool selection while paying insufficient attention to objectives, processes and capabilities. Technology should enable decision-making rather than drive the design of the measurement system itself. The most appropriate solution is therefore not necessarily the most sophisticated one, but the one that best supports the organization’s objectives and operating context.

Bringing the questions together

A recurring observation from the literature is that these dimensions influence one another. To illustrate this through the stockout scenario, many organizations would naturally focus on the What by introducing additional KPIs, dashboards or reporting mechanisms. However, stockouts rarely result from a single metric failure. Organizations may pursue ambitious objectives (Why) without possessing the necessary capabilities (Who) or technological infrastructure (Which) to support them, while performance measures (What) are often ineffective when visibility across the supply chain (Where) or measurement timing (When) are poorly designed. Consequently, stockouts frequently reflect a combination of issues relating to strategic intent, visibility, timing, organizational capabilities and enabling tools. The value of the 6Ws Framework is that it encourages managers to consider these dimensions collectively rather than in isolation.

Implications for supply chain managers

Define success before selecting measures. Too many organizations begin with KPI selection, but initially defining the desired outcome should ensure that performance measures support meaningful decision-making.

Extend visibility beyond traditional internal and external boundaries. Performance outcomes are affected by suppliers, internal operations and customers, and therefore, effective measurement systems require visibility beyond organizational boundaries.

Ensure technology supports the measurement approach rather than driving it. Technology should support performance measurement rather than define it, as even the most sophisticated dashboard will add little value if the objectives, capabilities and responsibilities remain unclear or are unsuitable.

Supply chain managers are not short of metrics, and there are a range of readily available and well-established approaches, such as SCOR and the Balanced Scorecard. The 6Ws Framework is not intended to replace them or diminish their usefulness. Instead, it provides a structure for thinking about how those approaches can be applied more coherently and effectively. It therefore provides an accessible and practical set of questions that can support the design, evaluation and implementation of supply chain performance measurement systems. Using this holistic approach to measurement system design can help organizations move beyond KPI selection, leading to better reporting, better decisions and ultimately better performance.

About the author

Dr. Stephen Kelly is a Senior Lecturer / Faculty of Social Sciences, Humanities and the Arts for People and the Economy (SHAPE) at the University of Salford. He can be reached at s.j.h.kelly@salford.ac.uk.

Today is Wednesday, Nov. 11.

1. AI capacity race tightens supplies of mature automotive chips

Semiconductor makers warn that foundries and packaging providers are redirecting resources toward higher-margin AI products, worsening shortages of mature-node components used in vehicles and industrial equipment after years of limited investment.

Reuters  |  Read the original article

2. China builds automated-port influence across global trade lanes

China is expanding automated terminals, digital trade documents and AI-enabled customs systems at home and abroad, raising efficiency prospects alongside Western concerns about surveillance, economic leverage and dependence on Chinese logistics technology.

Reuters  |  Read the original article

3. Honda commits $3 billion to a flexible Ohio vehicle plant

Honda plans an Ohio factory capable of producing hybrid, gasoline and battery-electric vehicles, adding up to 240,000 units of annual capacity and more than 2,000 jobs when production begins in 2029.

Supply Chain 24/7  |  Read the original article

4. EIA raises its 2027 diesel forecast as fuel supplies stay tight

The EIA now expects retail diesel to average $5.03 per gallon in 2027, 12% above its October forecast, as shipping disruptions and constrained global supplies support higher refined-product prices.

U.S. Energy Information Administration  |  Read the original article

5. Transportation inflation reshapes sourcing and mode decisions

Third-quarter executive commentary shows higher fuel and freight costs pressuring manufacturers while improving rail competitiveness and favoring refiners with shorter supply lanes, making network design a growing determinant of margin exposure.

FreightWaves  |  Read the original article

6. Autonomous truck economics could reach parity by 2028

A Kearney and MIT model projects U.S. driverless trucks could cost $2.60 per mile in 2028, but the crossover depends on fleet scale, technology costs, utilization and unresolved regulatory and liability questions.

FreightWaves  |  Read the original article

7. Stellantis sale shifts an idled Canadian plant toward defense vehicles

Stellantis agreed to sell its former Brampton assembly plant to armored-vehicle maker Roshel, drawing union objections that defense work cannot replace the automotive capacity and thousands of jobs previously supported there.

Transport Topics / Bloomberg  |  Read the original article

8. Cold-storage inventories stabilize after operators reduce capacity

Lineage said customer inventories have normalized at lower levels as the company and Americold cut 9 million square feet of capacity over two years, while higher fuel costs continued pressuring food volumes and transportation expense.

FreightWaves  |  Read the original article

9. Fleet operators signal another investment cycle

A Bain survey of 210 U.S. fleet operators found 79% expect their fleets to grow during the next year, with fuel prices, total ownership cost and changing technology influencing purchasing decisions.

Supply Chain 24/7  |  Read the original article

10. MSHA proposes revisions to miner silica protections

MSHA’s proposal retains existing silica exposure limits while revising monitoring, medical surveillance, recordkeeping and respiratory-protection requirements, with comments due Jan. 11 and a virtual public hearing scheduled for Dec. 15.

Work Safety 24/7  |  Read the original article

Executive takeaways

Editor’s note: This is Part 2 of a 4-part series on multi-tier supply chain networks. Part 3 in the series will publish on Tuesday, Nov. 17, with Part 4 publishing on Nov. 24. You can read Part 1 here.

The Critical Path Method (CPM) was developed in the late 1950s by Kelley, Walker, Malcolm and others, alongside PERT (Program Evaluation and Review Technique). It is a classic of project management, and what it works with is the lead time of each process (LT, in units of time/pcs). It traces every route that converges from a starting point—raw material to an end point to the finished product—through thousands of processes. Each of those routes is called a path, and the longest of them all is the critical path.

The founders of CPM and PERT fixed on that longest path for a reason: the completion time of the whole project is set by the route that takes longest. Its native setting, though, was the one-off large project—construction, systems development, a space program.

The Theory of Constraints (TOC), by contrast, is the newer of the two, put into the world by Eliyahu Goldratt in 1984 in his book The Goal. Its setting was the factory, repetitive production. What TOC works with is throughput—the rate of output. It matters which scope we mean. System throughput is the quantity of finished product the whole network turns out per unit of time; process throughput is the quantity a single process can handle per unit of time (both in pcs/time). In any system there is at least one node that is holding back the growth of total output. Where processes sit in a straight line, it is the node with the lowest value when process throughputs are compared, the process with the least capacity. Goldratt called it the bottleneck (BN), or the constraint. What TOC has studied ever since is how to find that BN, and what to do once you have.

Once the network branches and converges, though, it is not so simple: a node of low capacity can fall into the shadow of another and never surface. That is why, in my research, I gave every node its spec and ran material through a simulation to confirm the true BNs. How far that verification reached is what Part 4 of this series (publishing Nov. 24) takes up.

What struck me

CPM and TOC have rarely been brought onto a single scale. Part of the reason is that they were born in different settings—CPM in the one-off project, TOC in the factory floor of repetitive production. But there is a deeper reason: their units are reciprocals of each other. The LT that CPM works with is time per piece (time/pcs); the throughput TOC works with is pieces per unit time (pcs/time). The same phenomenon, with numerator and denominator exactly swapped. Goldratt’s own Critical Chain (CCPM) is the exception that brought TOC’s notion of the constraint onto a CPM network, but that was an integration premised on having every process duration already in hand.

So here is the thought: could the fact that the units are reciprocals be turned around and used as the handle? Could TOC’s question—where is the node with the lowest throughput?—be restated as CPM’s question: where is the longest path by LT?

If it can, the consequence is large. To compare the throughput of every process at every firm in the network directly, each firm would have to disclose the capacity of all its internal processes. And that, as we have seen, does not come back. But if that restatement is permitted, what is needed is no longer process-by-process throughput data, but only each firm’s company-level LT. Narrowed to that alone, might the BN across the whole network be found while disclosure is held to a minimum? That thought was the starting point for the idea of a platform to make BNs visible across an entire supply chain network (SCN).

BOM-Batch conversion

There is one obstacle to clear before the conversion works: processes count in completely different units.

A blast furnace runs three shifts around the clock and pours tens of tons of steel at once. A chemical plant, though it ships in bottles, handles solvents, reagents and paints in enormous tanks, a great volume at a time. A metal press, like a taiyaki mold (or a waffle iron), stamps out many screws or gears in a single stroke. And at the end of a car plant’s final line, vehicles come together one at a time.

Here is the problem. One car takes four tires, two headlights, one steering wheel. If a single batch yields eight at the tire plant, four at the light plant, and three at the steering wheel plant, then stringing together each process’s per-batch LT is simply wrong. Operating hours differ too: a blast furnace running 24 hours cannot be lined up against a screw plant working an eight-hour day. Machine maintenance intervals differ; so do defect rates.

So every process is converted into one common unit: one finished vehicle. Each process’s recorded per-batch LT is aligned to the pace at which the car plant turns out one vehicle, and converted according to the quantity of each part one car requires, as listed in the bill of materials (BOM)—four for tires, one for the steering wheel—then laid out again. Differences in operating hours from shift patterns and maintenance stoppages are folded into the conversion, and so is the defect rate: time spent producing defective parts is, in effect, time the process stood idle.

Only then does every process across the whole SCN line up on the same ruler. The quantity of interest is this: how many minutes does one process inside a supplier take to finish the quantity of parts that one finished vehicle requires? After conversion, the larger a process’s LT, the less it can produce per unit of time—which is to say, the lower its throughput capacity. I have called this operation BOM-Batch conversion. In implementation, it is desirable that each figure carry not only a mean but attributes such as a standard deviation, a sample size, and the period over which it was compiled.

This conversion is the function at the core of what an information-sharing platform should provide. The work of getting each firm’s numbers onto the common ruler is taken on by the platform, on the condition that nothing is exposed to other firms.

What company-level lead time is

So what, concretely, is the “single number” each firm hands over? There are at least two ways to measure it.

As an approximation, the inbound timestamp and the outbound timestamp—the difference between those two moments will do. It is the time between the moment material passes through the inbound gate of the factory and the moment product leaves through the outbound gate. How much LT each individual process in between consumed is not asked at all. For most firms, those two times are already somewhere in their records.

Read Part 1: Projecting constraints in multi-tier supply chain networks: Why suppliers keep their process data closed

For a more accurate figure, sum the BOM-Batch-converted LTs of the individual processes along the internal route. In doing so, the time work-in-process (WIP) spends waiting between processes is also counted, treated as a process in its own right—the work of carrying a part and handing it on. How that waiting time should be handled is something I will return to in Part 4. For a firm that knows its own process breakdown, this method is the more precise of the two.

One clarification is needed here. Some factories are a straight line: a single part arrives, successive machining steps are applied, and eventually it ships. But in a factory where several materials are brought together into an assembly, the internal process flow also branches and converges like a river. In that case, what counts as the company-level LT is the longest of those converging internal routes by total LT.

Where the platform’s bottleneck search comes to a halt

Here is the design point on which the whole method turns. What the platform identifies from outside is at the granularity of the firm. Up to the point of identifying which firm on the critical path has the longest company-level LT, everything can be calculated from the single number each firm has handed over. Inside the factory, the platform’s BN search stops; it does not, and need not, go looking there.

From there, the road divides in two.

One case is the firm that does not itself know where its own constraint lies. The process-by-process data is not there; it cannot describe its internal structure in numbers. Here the platform takes over the search, in the following way. It splits the firm’s internal processes into an upstream group and a downstream group at the point that divides their total LT in half, asks which group is the longer, and splits that group again the same way. Narrowing this way until no further split is possible will reach the constraint, but along the way some information about processes that have nothing to do with the constraint also leaks out as collateral. The leakage is an unavoidable cost—but of the disclosure strategies I compared, this midpoint split keeps the cost lowest, reaching the BN in the fewest visits while leaking the least. And unlike the two simulations I turn to later in the series, this result rests on statistical ground: it held across many randomized networks, under two input distributions and three network shapes. It was the first of my dissertation’s three simulations.

The other case is the firm that, through digital transformation (DX), does know where its own constraint is. Here the platform does exactly one thing. It asks that firm: regarding the constraining process in your plant, the throughput capacity alone will do—could you tell us? Which process that is, the firm decides for itself. And it discloses only the specification of that constraining process—it never goes through the split just described at all. No collateral disclosure about any other process occurs.

Put another way: the better a firm understands its own interior, the less of it needs to be opened. The effort of making one’s processes visible is itself the power to keep internal information protected. The platform’s BN search stops at the factory gate; beyond it, the firm guides the way.

And consider what that means across the whole process. Every participating firm in the network but one was asked for a single thing: its company-level LT. Only the one firm projected to hold the network’s BN was asked anything further—and even then, only for the specification of that one constraining process.

For the BN search, that is the whole of what the platform asks. Say the network’s BN has been drawn out this way, what happens then? Next time, I want to follow what improving it sets in motion, and what the supplier, the original equipment manufacturer (OEM), and the bank each stand to gain.

About the author

Brian (Yoshiatsu) Kawabata began his career at Honda, earned an MBA at Toulouse Business School in 2008, and went on to work at Boeing and Intel. In June 2026, he completed a PhD in computer science at Tokyo University of Agriculture and Technology, with a dissertation on detecting the constraints of multi-tier supply chain networks from minimal shared information; a related paper appears in Information Technology and Management (Springer, 2026; https://link.springer.com/article/10.1007/s10799-025-00478-x). He writes from both sides of the work—the factory shop floor and the theory.

Today is Tuesday, Nov. 10.

1. Supply disruption forces shift changes at three BMW plants

BMW adjusted shift patterns at three plants after supply problems interrupted production, while the automaker worked with suppliers to resolve an issue whose affected components, locations and duration were not disclosed.

Reuters  |  Read the original article

2. Federal agencies warn that hackers are targeting cold-chain controls

A joint federal alert says attackers are seeking remote access to refrigeration controllers, creating risks to food, medicine and equipment while prompting operators to isolate systems, patch firmware and strengthen authentication.

Transport Topics  |  Read the original article

3. Extended peak gives way to slower U.S. container imports

Global Port Tracker expects major U.S. ports to handle less import volume through year-end after September reached 2.37 million TEUs, as retailers finish holiday positioning and West Coast rail delays persist.

Supply Chain 24/7  |  Read the original article

4. FCC clears selected imported warehouse robots under new controls

Several foreign-made mobile robots and automated material-handling systems received conditional FCC approval, providing a limited import path as new equipment-authorization rules restrict advanced robotic devices deemed a national-security risk.

Supply Chain 24/7  |  Read the original article

5. Bangkok cargo-system failure creates a manufacturing bottleneck

A warehouse-management malfunction at Bangkok’s main airport is delaying cargo release, prompting inbound restrictions and freighter suspensions while manufacturers and forwarders face longer waits for electronics, medical and other time-sensitive shipments.

FreightWaves  |  Read the original article

6. U.S. Steel labor talks raise the risk of a production stoppage

United Steelworkers negotiators warned talks with U.S. Steel are unlikely to extend into 2027, as the parties continue renewing an expired contract covering more than 10,000 workers in 48-hour increments.

Transport Topics / Bloomberg  |  Read the original article

7. Roadside inspections intensify carrier compliance pressure

Federal and state inspectors completed more than 1,000 truck and bus inspections in Maryland and Washington, placing 92 Maryland drivers and 191 vehicles out of service while identifying additional violations in the District.

FreightWaves  |  Read the original article

8. McDonald’s gives Uber priority for app-order delivery

McDonald’s will prioritize Uber for orders placed through its app in 14 countries while remaining available on DoorDash, expanding a white-label delivery relationship that reaches more than 45 million U.S. loyalty members.

Transport Topics / Bloomberg  |  Read the original article

9. Wholesale inventories rise as sales reduce the stock-to-sales ratio

U.S. wholesale inventories increased 0.7% in September and 7.6% year over year, while sales grew 2.8%, reducing the inventory-to-sales ratio to 1.21 months from 1.23 in August.

Reuters  |  Read the original article

10. Forward Air pauses intermodal sale as performance improves

Forward Air halted the planned divestiture of its intermodal unit after management said operations returned to target performance, although the carrier remains willing to consider offers that reflect its assessment of fair value.

FreightWaves  |  Read the original article

Executive takeaways

Enterprise technology has never moved faster.

Many organizations have spent the last several years modernizing their digital core, moving critical business processes to the cloud, connecting fragmented systems, and strengthening data and analytics capabilities. Now, while some of those transformations are still maturing, leaders are being asked to prepare for the next wave: generative AI, intelligent automation, and autonomous agents.

The possibilities are significant, particularly across supply chain and manufacturing. AI can help organizations anticipate disruptions, improve planning decisions, accelerate exception management, automate repetitive activities, and increasingly take action across connected business processes.

But before asking what AI can do next, leaders should ask a more fundamental question: Did our previous transformation actually solve the problems it was intended to solve?

Digital transformations usually begin with recognizable business problems: too much manual work, fragmented systems, inconsistent data, slow decisions, excessive reconciliation, and legacy applications that are expensive and difficult to maintain.

New technology is introduced to solve those problems and create business value.

Then another technology wave arrives.

The terminology changes—from digital transformation to cloud, from cloud to advanced analytics, and now to generative AI and autonomous agents. Yet many of the problems being discussed remain familiar.

If each new technology wave is being asked to address the same underlying issues, the constraint may no longer be technology.

It may be the organization’s ability to absorb transformation.

Every transformation should pass forward stronger capabilities, making the next transformation easier—not forcing the organization to start over.

Adoption is only the beginning

Enterprises have become increasingly capable of implementing technology. Cloud platforms enable organizations to introduce new capabilities more continuously, while AI is increasingly embedded directly into enterprise applications. But technical availability does not automatically translate into operational value.

An organization can successfully implement cloud ERP and still depend heavily on spreadsheets and manual workarounds. It can deploy sophisticated planning capabilities while important decisions continue to happen outside the system. It can introduce AI-generated recommendations without clearly defining who should trust, challenge, approve, or act on them.

Technically, the capability has been adopted. Operationally, it has not been fully absorbed.

Adoption is the ability to acquire and implement technology. Operational absorption is the ability to integrate that technology into processes, decisions, data, governance, and ways of working so that it becomes part of how the business operates.

This distinction matters because technology cycles are compressing. Organizations can no longer assume they will complete one transformation, operate in a stable environment for several years, and then prepare for the next.

New capabilities are arriving while earlier transformations are still being stabilized.

The challenge, therefore, is not simply becoming better at implementing technology. It is becoming better at continuously absorbing change.

Why do the same problems keep returning?

Many transformation barriers are not purely technology problems.

They may originate in poorly designed processes, inconsistent data, unclear ownership, unnecessary customization, fragmented decision-making, limited understanding of end-to-end business processes, or a culture that allows temporary workarounds to become permanent.

Technology can expose these issues and help address them, but technology alone rarely eliminates them.

This becomes particularly important as organizations move toward AI and autonomous agents.

If a business process remains unnecessarily complex, automating it does not necessarily make it better. If underlying data cannot be trusted, a more sophisticated algorithm does not automatically produce a better decision. If decision rights are unclear, adding an AI recommendation may introduce another layer of uncertainty.

Before investing in the next technology, leaders should therefore ask: Are we using new technology to solve a new business problem, or are we using it to compensate for an old problem we never fully solved?

The hidden constraint: Operating model debt

Unresolved transformation issues accumulate over time.

A business exception leads to a customization. A process gap creates a spreadsheet. An urgent requirement results in a point-to-point integration. A temporary approval becomes permanent. Different functions develop their own processes and definitions.

Each decision may be reasonable in isolation. Collectively, they create operating model debt.

Like technical debt, operating model debt makes future change harder. Every new capability must navigate existing process complexity, organizational boundaries, inconsistent data, custom integrations, unclear ownership, and established workarounds.

This matters enormously for AI.

An AI agent working across planning, procurement, manufacturing, inventory, maintenance, or logistics depends on the processes, data, decision rules, integrations, and governance underneath it. If those foundations are weak, adding intelligence at the top does not automatically strengthen them.

In fact, greater automation can amplify weaknesses because decisions and actions may happen faster and with less human intervention.

The next generation of transformation, therefore, cannot focus only on adding capabilities. It must also remove the organizational friction that prevents those capabilities from creating value.

Start With the business problem, not the technology

As emerging technologies gain attention, it is easy to reverse the sequence of transformation: organizations see a new capability and then begin looking for places to use it.

A stronger approach starts with the business.

Every investment in technology should begin with two questions:

1. What business problem are we trying to solve?
2. What measurable value will it deliver?

This applies whether an organization is replacing legacy ERP, introducing cloud planning, implementing advanced analytics, deploying generative AI, or experimenting with autonomous agents.

But as organizations move from one technology generation to another, a third question becomes equally important: Why are we still solving this problem?

If a cloud transformation was intended to eliminate fragmented processes, manual reconciliation, and inconsistent information, an AI initiative should ideally build on that foundation to address higher-value challenges.

AI might help anticipate supply risks earlier, identify patterns that are difficult for people to detect, improve planning decisions, orchestrate responses across functions, or autonomously manage well-defined exceptions.

It should not be necessary to use AI simply to compensate for fragmentation that the previous transformation was expected to eliminate.

Each technology wave should increase what the organization is capable of doing. That is the difference between accumulating technology and building business capability.

From speed to deployment to speed to value

Transformation programs have traditionally focused heavily on implementation milestones: How quickly can we move to the cloud? When can we migrate users? When can we go live?

Those questions still matter, but go-live is the beginning of value realization, not the end.

The more useful measure is speed to value: how quickly the organization can move from technical availability to sustained business use and measurable outcomes.

Leaders can evaluate this through measures such as:

• Time from deployment to stable business operation
• Percentage of transactions following standardized processes
• Reduction in manual workarounds and exceptions requiring human intervention
• Time required to integrate new capabilities into core processes
• Measurable business value realized after deployment

These measures provide a different view of transformation success. A project may have been delivered on time and on budget. But if the organization continues to rely on the same manual interventions, workarounds, and fragmented processes, the technology may have changed without sufficiently changing how the business operates.

Transformation should therefore accomplish more than implementing a new platform. It should solve meaningful business problems, create measurable value, remove unnecessary complexity, and increase the organization's capacity for future change.

Every transformation should make the next one easier.

Building operational absorption capacity

If technology cycles continue to accelerate, organizations cannot afford to rebuild their operating model around every new wave.

This essential organizational capability is best understood as operational absorption capacity: the ability to introduce new capabilities, integrate them into the business, create value from them, and move to the next opportunity without repeatedly reopening foundational problems.

Three shifts can help build that capacity.

1. Modernize the digital core without recreating legacy complexity

Moving to the cloud should simplify the enterprise, not reproduce years of legacy complexity on a newer platform.

That requires discipline around process standardization, customization, integration, data architecture, and solution design.

Every customization should be evaluated not only by the immediate requirement it solves but also by the future complexity it creates. The same thinking should apply to integrations and process exceptions.

For supply chain organizations, ERP, planning, manufacturing, logistics, maintenance, and data platforms should increasingly operate as parts of a connected digital core rather than as independent technology projects.

A strong digital core should make it easier to introduce new capabilities without repeatedly rebuilding the foundation underneath them.

2. Move from better information to better decisions

Traditional enterprise systems largely focused on recording transactions. Analytics improved visibility. AI increasingly changes how decisions themselves are made.

This requires organizations to think beyond dashboards and algorithms.

Which decisions should remain human? Which should be augmented by AI? Which routine decisions can eventually be automated? When should an autonomous system escalate an exception? What level of confidence is required before a recommendation becomes an action? And who remains accountable for the outcome?

These are not simply technology questions. They are operating-model questions.

Organizations that establish clear decision rights and governance will be better positioned to absorb increasingly autonomous capabilities without reinventing their controls each time technology advances.

3. Build capabilities that outlive the technology

Specific technologies will continue to change. The organization’s ability to change should endure.

Employees do not need to become experts in every emerging AI technology. Organizations do need people who understand business processes end to end, can interpret data, challenge recommendations, redesign workflows, manage exceptions, collaborate across functions, and continuously learn.

The same principle applies to governance.

Instead of creating a new governance model for every technology wave, enterprises need repeatable mechanisms for evaluating business value, managing risk, determining appropriate human oversight, measuring adoption, and retiring processes or capabilities that no longer add value.

The objective is not simply to become ready for AI. It is to build an organization capable of absorbing whatever meaningful technology comes next.

A different executive conversation

As AI becomes embedded throughout enterprise technology, leadership teams should expand the questions they ask.

Instead of beginning with, What new AI capabilities should we implement?, leaders should ask:

• What problem are we trying to solve, and what measurable value will solving it create? 
• Why are we still solving this problem? 
• Are we automating a well-designed process or automating around an unresolved process problem? 
• Can we adopt new capabilities without extensive integration and process redesign? 
• Did the last transformation increase our ability to absorb the next one?

These questions shift the conversation from technology adoption toward business capability and organizational progress.

The next transformation should be easier

There will always be another technology wave.

Today, the discussion centers on generative AI and autonomous agents. Tomorrow, enterprises will be evaluating capabilities that are only beginning to emerge.

Organizations cannot predict every technology that will matter five years from now, nor should they attempt to adopt everything that appears. What they can do is ensure that each transformation leaves the enterprise stronger and better prepared for the next one.

A successful digital transformation should establish a stronger digital core. That foundation should make advanced analytics and AI easier to adopt. AI, in turn, should enable organizations to address more complex decisions, automate well-defined processes, and pursue opportunities that were previously difficult to reach.

Each wave should move the organization forward.

If the same fundamental problems continue to reappear, leaders should understand why the previous transformation did not fully resolve them before adding another layer of technology.

The real measure of transformation is not simply whether a new technology was successfully implemented. It is whether the organization converted that technology into business value while strengthening its ability to absorb what comes next.

The enterprises best prepared for the next era will not necessarily be those that adopt every emerging technology first. They will be those that continually strengthen their processes, data, governance, decision-making, and workforce as technology evolves.

Because ultimately, every transformation should solve meaningful business problems, create measurable value, and make the next transformation easier.

About the author

Srinivasan Narayanan is an Oracle Cloud Solution Delivery Lead with extensive experience supporting large manufacturing organizations through ERP-driven supply chain, manufacturing, and maintenance transformations across multi-plant and multi-distribution-center environments. He is an IEEE senior member, industry speaker, and contributor to practitioner-focused research and trade publications on ERP modernization and operational resilience.

Executive takeaways

The stunning results of the midterm elections are likely to have an effect on global supply chain operations. With Democrats now holding comfortable majorities in both houses of Congress and Trump still in the White House, it could lead to gridlock in Washington. Trump has already said he would become ”Mr. Veto,” meaning he would block any legislation the Democrats advance. But the House and Senate majorities could also lead to a softening in Trump’s draconian trade policies and restrictions on funding for the Iran war.

Here is what we should consider as we determine the path ahead for our global supply chains.

A more stable economy

A research study conducted by the Reshoring Institute in early 2025 indicated that manufacturers across America were holding back on making capital investments and hiring due to the instability in the economy and volatility in tariffs.  The one thing that manufacturers hate is unpredictability. Not being able to make assumptions about the future makes supply chain and manufacturing planning very difficult. Companies were reacting to this instability by doing nothing or very little investing, building, hiring, or reshoring.

We know from our research that a magnitude of capital investment dollars have been sitting on the sidelines until tariffs, trade issues, and the war in Iran stabilize or have a pathway for resolution. A somewhat more stable and predictable economy is likely ahead with the new Congress, even if this brings stalemates between Congress and the White House. At the very least, stalemates are somewhat predictable.

Capital investment

It appears that there is a lot of pent-up demand for manufacturing expansion and reshoring simply because there has been so little investment over the past two years. Hiring in the manufacturing sector has been flat according to the U.S. Bureau of Labor Statistics since the beginning of 2025. A more stable economy, even if the tariffs continue, will likely free up the waiting investment capital, which should result in new manufacturing facilities and more jobs created.

Tariffs and trade

Over the past two years, President Trump has imposed various tariffs through the use of emergency powers and signing Executive Orders. To date, most of these tariffs have been declared illegal by the courts and duties collected are now being refunded to importers. The authority for imposing tariffs and taxes rests with Congress under the Constitution, and it is likely that the new House of Representatives will fight to take back that responsibility. Tariffs have been very unpopular with the American public because they increase prices for consumers, and while additional tariffs are possible under new Executive Orders by Trump, there will likely be less focus in this area.

Trade and foreign policy remain the responsibility of the Executive Branch. Because supply chains are global in nature, and trade is reliant on good trade relationships, this too is an area likely to change. Congressional delegations are likely to be very active in traveling to U.S. trading partner countries to rebuild bonds and improve relations. This is good news for reducing friction in global trade. 

Taiwan

Trade with Taiwan, where most of our advanced semiconductors are fabricated, remains in question. U.S. support for the Taiwanese government is a problem for our relationship with China, with the White House appearing to favor China. The size, strength, and U.S.-China market integration must also be considered. The White House policy on China is unclear, while  Congress, over the past few years, has favored Taiwan. This is a wait-and-see topic.

USMCA

The USMCA agreement was supposed to have been updated in July, but the Trump administration refused to sign the continuing proposal, preferring instead to conduct bilateral negotiations individually with Canada and Mexico. Lobbying by the automotive industry with Trump and Congress will likely result in favorable terms for trade with Mexico, where the automotive industry is heavily invested.

Canada is a different story. Trump’s feud with Canada will likely be affected by the U.S.-Canadian border states of Michigan, Minnesota, Ohio, Maine, and New York, where newly elected Democratic representatives and senators are likely to put pressure on Trump to back off on the Executive Order tariffs and make peace with Canada.

Data centers and AI development

This issue lands more on the states than Congress. Beginning in 2027, the U.S. will have 29 elected Democratic governors. Many of the newly elected governors have promised to slow or stop the construction of data centers until reasonable solutions are found and agreements can be made regarding electricity, water, and noise pollution. Electricity and water are also used in abundance by manufacturing sites. Pauses to new data center construction and new solutions for energy and water may bring beneficial changes for everyone. Look for innovative energy generation, water recycling, and improvements in infrastructure to support both data centers and manufacturing.

Iran war

Gas and diesel prices were at the forefront of most of the election campaigning, with the war on Iran blamed for increases. According to the U.S. Global Leadership Coalition, a staggering 74% of voters called the Iran war a major issue in their vote. The election results will most likely bring a speedier end to the war through funding cuts and pressure on the White House, and a resumption of the oil trade in the Middle East.

A reduction in gas and diesel would bring relief to consumers and bring down transportation costs. It will likely be a while, though—maybe a year or more—before we see the effects.

Change ahead

Change is on the horizon and on the agenda for most companies. Supply chains are likely to benefit from the road ahead through innovative solutions and improved international relations.

Today is Monday, Nov. 9.

1. China extends its U.S. trade truce for two months

China will continue suspending a 24% additional tariff on U.S. imports while retaining a 10% levy, giving negotiators two more months to pursue a broader reciprocal trade package.

Reuters  |  Read the original article

2. Southern California ports gain two more weeks to avert a strike

Port employers and union watchmen extended contract talks through Nov. 20, averting an immediate walkout by workers whose picket line could interrupt operations at Los Angeles and Long Beach terminals.

Reuters  |  Read the original article

3. High tanker costs shift refining economics toward the U.S. Gulf Coast

Hormuz disruptions have lifted tanker and insurance costs, widening discounts on nearby Canadian and Venezuelan heavy crude and giving U.S. Gulf Coast refiners an advantage over Asian competitors.

Reuters  |  Read the original article

4. USMCA uncertainty pushes North American freight networks toward flexibility

Shippers and carriers are adding redundant capacity, multimodal options, flexible warehousing and customs expertise as annual USMCA reviews, shifting tariffs and tighter truck capacity complicate long-term network decisions.

Transport Topics  |  Read the original article

5. Component shortages constrain a rapid truck-production recovery

Axle and engine shortages limited Daimler Truck North America output and increased expedited-freight costs during the third quarter, with supply constraints expected to continue through the fourth quarter.

Transport Topics  |  Read the original article

6. Compromised communications drive a large share of reported load thefts

Highway attributed 40.2% of third-quarter stolen-load reports to compromised communications, while fraudulent calls rose 90.3% year over year and carrier ownership-change reports increased sharply during the period.

FreightWaves  |  Read the original article

7. Walmart and Zipline plan a nationwide drone-delivery expansion

Walmart will add Zipline drone service at 300 stores as the operator raises $1.5 billion for manufacturing and expansion, testing how store proximity can support faster last-mile fulfillment.

FreightWaves  |  Read the original article

8. North American container imports grow as freight prices accelerate

North American container imports increased 9% year over year in September while a global price index rose 62%, showing resilient cargo demand alongside disruption, capacity limits and higher operating costs.

FreightWaves  |  Read the original article

9. Union Pacific rate comments intensify scrutiny of proposed rail merger

Shipper groups raised retaliation concerns after a Union Pacific executive’s rate remarks became public, adding a transparency issue to regulators’ review of the railroad’s proposed Norfolk Southern combination.

Supply Chain Dive  |  Read the original article

10. AI-generated evidence raises the verification burden in supplier audits

Generative AI can fabricate mutually consistent documents, images and testimony, prompting supply-chain auditors to strengthen provenance checks, independent confirmation, physical inspection and escalation rules for high-risk claims.

Supply Chain Management Review  |  Read the original article

Executive takeaways

Supply chain audits rest on a core assumption: compliance can be verified by checking evidence. Non-financial evidence can come in many forms, such as internal reports, staff interviews, or sensor readings. Fabricating evidence was once difficult, time-consuming, and expensive. However, generative artificial intelligence can change that. It can now produce both legitimate assistance and fraudulent evidence, and auditors need a reliable way to tell the two apart.

A supply chain audit runs as a chain of evidence, from sampling and inspection through review to attestation and surveillance. AI can attack any link in this chain and can quickly generate many pieces of evidence that appear to corroborate one another. A supplier faking organic cotton, for example, may have once had to manually forge transactions and painstakingly build a fraudulent verification channel. AI doesn’t change the motivation for fraud, but the "filter of effort" that once limited it to skilled actors has collapsed.

The potential use of AI in fraudulent activities is increasingly recognized, and guidance for auditors on AI-generated evidence is emerging. Much of the focus to date, however, has been on financial auditing, and not the broader quality, social and environmental, or health & safety issues that are the focus of many supply chain audits. Detection of AI-fabricated content is becoming urgent, and the following is a warning about structural vulnerabilities and how they can be addressed.  

Where evidence can break down

Expectations for evidence in supply chain audits generally emphasize characteristics such as veracity, timeliness, relevance, reliability, and independence. Vulnerabilities occur across all types of audit evidence: documentary, visual, testimonial, and physical. While not all use of AI is fraudulent, such as in auto-drafting summaries of real meetings or supporting data analysis, it can also create new opportunities for deception.

Documents are the clearest vulnerability, and supply chain auditing experience with this form of fraud pre-dates AI. Kobe Steel, for example, falsified data on the quality of its products for decades before being caught in 2017. AI’s significance is, therefore, not always in new forms of fraud, but in lowering the cost of building convincing evidence. AppZen, an expense auditing company, has flagged the growing risk of AI-generated receipts to support expense fraud. Although expense receipts are not necessarily supply chain records, the underlying vulnerability is similar: AI can rapidly construct plausible evidence for transactions that never occurred.

Visual evidence, including photographs and video, was long viewed as proof of physical reality. Video cameras have already been used to capture information on fishing catch and location, and there are emerging efforts to use AI to improve this. The rapidly improving ability of generative AI to generate or alter visual evidence, however, creates new opportunities for manipulation. Establishing provenance through metadata, geolocation, timestamps, and, critically, physical inspection will take on new importance in future supply chain audits.

Increasingly, synthetic voices, faces, and video can raise concerns regarding the veracity of testimonial evidence. While it has not yet been documented in a supply chain audit context, deepfake videos are already being used for fraudulent business purposes. Even genuine human participation is not free from the potential influence of AI. Although the research on its current effectiveness is mixed, real-time AI coaching can guide those that do participate. Independently recruited worker interviews, conducted without management present and backed by identity and location checks will become increasingly valuable.

Physical evidence also has vulnerabilities to AI-generated fraud. Sensors can be compromised, measurements can be selectively reported, and remote data can be misread or manipulated. These forms of evidence, however, can often be gathered independent from the organization’s own reporting systems, which makes them harder to fabricate at scale. We have conducted extensive research on the use of remote auditing technologies, which has focused on strengthening exactly this kind of evidence and how it complements traditional auditing.

How to build trust in audits in the age of AI

AI can generate entirely synthetic evidence with no authentic original, obscure traditional cues for detecting forgery, and manipulate evidence in ways that are hard to detect. The old question—does this evidence appear credible?—is no longer sufficient because generative AI can produce a complete chain of evidence that is internally consistent but partially or completely fabricated.

Supply chain audits must vigilantly guard against conflating four things: where a piece of evidence came from (i.e., origin), whether it has been altered (integrity), who stands behind it (identity), and whether what it represents actually happened (truth). Verifying that a supplier met a standard, or a process occurred, now depends more than ever on a connection to the physical world. No single source of evidence, no matter how convincing, should stand alone.

This brings supply chain auditors under the spotlight. In the new AI-driven business environment, auditors must re-focus on confirming evidence at the source. This could include direct confirmations with external parties and unannounced visits. They also need to recognize that internal consistency does not mean independence. For example, records generated or held by a single party may agree with each other even if the underlying claim is untrue. Auditors must thus test the digital chain of evidence, as well as the physical reality it claims to reflect.

A healthy skepticism remains the essential starting point for weighing key evidence. Unusual metadata or excessive uniformity should be investigated, but do not on their own indicate fraud. Auditors should treat suspicion of AI use as something to investigate. This mirrors an approach long used in journalism, where reporters verify a claim by tracing it to its origin and deliberately seeking disconfirming evidence.

Moving forward

Since the COVID-19 pandemic, senior supply chain leaders have been under pressure to innovate supply chain audits and cut audit costs, leading to greater use of remote audits and mixing them with traditional on-site audits in hybrid forms.

Much of our advice focuses on embedding audits in the physical world. Yet on-the-ground verification and manual review of evidence are expensive, running against the cost cutting trends behind remote audits. As AI narrows the gap between forged and genuine evidence, human judgement matters more than ever. This underlines that plausible-looking claims are no longer a reliable proxy for those that are genuine, whether they are investigated remotely or in person. Controls must be in place, including prohibited AI uses, evidence-retention and disclosure rules, enhanced procedures for high-risk claims, and consistent escalation pathways. Resourcing audits to apply these controls will be essential.

In part, organizational leaders can turn to external bodies for guidance. Global standards bodies such as ISO have already produced guidelines on auditing in AI-driven environments, as well as auditor competency standards. Rather than leaving individual auditors to determine how to evaluate AI-generated evidence on a case-by-case basis, supply chain leaders should issue supply chain-wide guidance on what counts as trustworthy evidence in an age of generative AI.

The underlying question in supply chain audits has not changed: can supplier claims be trusted? AI does not create motivations to deceive, but it does expand capabilities for those so inclined. In doing so, it erodes past assumptions regarding the cost, difficulty, and imperfections of fabrication. Answering questions of trust, therefore, requires auditors to look past evidence that appears convincing at first glance. AI has collapsed the filter of effort that once made fabrication costly. Audits demanding robust verification of evidence are the only way to rebuild trust.

About the authors

Cory Searcy is the Assistant Vice President, International and Professor of industrial engineering at Toronto Metropolitan University. His research focuses on supply chains, with an emphasis on sustainability, auditing, and reporting. He was a section editor at the Journal of Business Ethics for more than 10 years.

Pavel Castka is Professor at Xi’an Jiaotong‑Liverpool University, China. He has served for over two decades on committees at the International Organization for Standardization (ISO), including the development of ISO 26000 on social responsibility and ISO 9001 Auditing Practices Group. His work focuses on the innovation and digitalization of supplier audits.

Disclosure: Prof Castka is a member of the ISO 9001 Auditing Practices Group (ISO/TC 176), and guidelines produced by this group are referenced in this article. He serves in a voluntary capacity and receives no remuneration for this work.

Today is Friday, Nov. 6.

1. West Coast port labor talks move toward a new strike deadline

Federal mediation continued after port watchmen and terminal employers extended their contract through Friday, delaying a possible walkout that could disrupt cargo flows at Los Angeles and Long Beach.

FreightWaves  |  Read the original article

2. Higher U.S. tariffs threaten another contraction in Canadian cabinet production

Canadian cabinet makers are cutting output and investment as a 25% U.S. duty strains an export-dependent industry, with the rate scheduled to double to 50% on Jan. 1.

Reuters  |  Read the original article

3. Energy access emerges as a constraint on network and automation plans

Gartner expects grid congestion, electrification and data-center demand to make energy availability a decisive factor in site selection, network design and automation investments for manufacturers through 2029.

Gartner  |  Read the original article

4. Federal inquiry puts containerboard pricing under scrutiny

Smurfit Westrock, International Paper and Packaging Corporation of America disclosed grand jury subpoenas in a federal investigation examining possible price fixing in containerboard and related products.

Transport Topics / Bloomberg News  |  Read the original article

5. Humanoid robotics listing brings warehouse automation economics into focus

Agility Robotics began public trading after raising $465 million through a SPAC transaction, pairing more than $300 million in conditional orders with substantial losses and limited first-half sales.

MarketWatch  |  Read the original article

6. Capacity exits start to lift trucking economics despite flat freight demand

ATA economist Bob Costello said trucking has entered an early supply-driven recovery as capacity contracts, citing a 32% year-over-year increase in third-quarter shipper spending while shipment volumes stayed largely flat.

Transport Topics  |  Read the original article

7. Cargo thieves in Mexico increasingly target parked trucks

Parked vehicles accounted for 42.2% of reported Mexican cargo thefts in the third quarter, up nine percentage points, while 61% of incidents occurred between 6 p.m. and 6 a.m.

FreightWaves  |  Read the original article

8. Maersk lifts its outlook as ocean earnings outpace expectations

Maersk raised its full-year profit outlook after third-quarter EBITDA reached $4.7 billion, as stronger ocean rates offset slightly lower volumes and logistics revenue increased 13% year over year.

FreightWaves  |  Read the original article

9. Commodity visibility gives Unilever room to accelerate pricing

Unilever raised third-quarter prices 2.3% overall after improved visibility into crude oil and other input supplies, using pricing and pack architecture to offset elevated commodity and logistics costs.

Supply Chain Dive  |  Read the original article

10. Supplier simulations show when responsiveness can outweigh purchase price

An AI-assisted retail simulation found that rising forecast error and lead-time variability can reverse the economics of imported and domestic suppliers, supporting sourcing recommendations built around trigger conditions.

Supply Chain Management Review  |  Read the original article

Executive takeaways

Strategic sourcing has long been organized around a seemingly straightforward objective: identify the “best” supplier. Unfortunately, the answer may be valid only under a narrow set of assumptions. In an uncertain operating environment, executives need a different question: not simply which supplier is best, but when that answer changes.

The distinction is more than semantics. To explore those questions, we used generative AI to build a simulation model that evaluated supplier decisions across thousands of plausible operating environments. The results showed that even modest changes in demand or supply conditions were often enough to reverse the preferred supplier recommendation. The same technology behind our analysis now puts techniques once reserved for modeling specialists into the hands of sourcing managers without formal training in simulation or advanced analytics.

Built for a single future

For decades, strategic sourcing has become progressively more sophisticated. Companies moved beyond purchase price to total landed cost and cost to serve. Each iteration evolved to produce better decisions by capturing costs that had previously been ignored. Yet despite these advances, one assumption has remained largely unchanged: sourcing decisions are often optimized around a single expected future. Demand is forecasted, lead times are estimated, transportation costs are projected, and supplier reliability is assumed at average levels of performance. The supplier recommendation that emerges is only as good as those assumptions.

That approach was logical when analytical capability was limited and operating conditions were more steady-state. It makes far less sense today. Supply chains operate in environments where demand shifts unexpectedly, transportation networks tighten, suppliers miss commitments, competitors stumble, and geopolitical events reshape sourcing economics with little warning. Given increasing uncertainty, sourcing efforts need to shift from which supplier has the lowest cost to when does a supplier stop being the best choice.

That is a fundamentally different sourcing question, and the distinction matters because most supplier recommendations are remarkably sensitive to variation. A supplier can dominate a traditional sourcing analysis because it performs best under one set of normal and average conditions. However, simulations often show that if those conditions change only modestly, the preferred supplier can change with them.

Historically, there have been very few practical alternatives. Exploring vast combinations of various costs, forecasts, and performance indicators required specialized software, advanced modeling expertise, extended time, and considerable cost. Those resources were justified for a major network redesign but rarely for an individual supplier selection decision.

The new AI multiplier: From task to capability

Most discussions of AI in sourcing focus on automation. AI reviews contracts, summarizes proposals, compares supplier bids, and accelerates routine work, making it a powerful task multiplier. But we found that AI’s larger contribution is different. AI is becoming a capability multiplier by enabling sourcing managers to use sophisticated analytical methods, such as simulation and sensitivity analysis, that were previously practical only for specialized analysts. Rather than simply helping managers work faster, AI expands the kinds of supplier scenarios they are capable of evaluating.

Traditional total landed cost optimization seeks the single best answer. Simulation instead identifies the factors supporting a recommendation, the conditions under which another supplier becomes preferable, and the variables executives should monitor after the contract is signed.

Consider one of the most common sourcing decisions retailers make: choosing between a domestic and an imported supplier. Total landed cost almost always favors the import but rarely accounts for what happens when demand or lead times drift from plan. The sophistication of AI as a capability multiplier makes that evaluation practical for routine sourcing decisions.

What this means for every sourcing decision

Just as important as what the simulation revealed is how we developed it. In our analysis, generative AI helped translate the retailer’s replenishment rules, cost structure, and supplier performance assumptions into a working simulation model. Rather than manually programming every element from scratch, we could describe how the sourcing system operated, refine the model through successive interactions with AI, and evaluate thousands of combinations of demand and lead-time uncertainty. The underlying analytical techniques are not new. What has changed is the ability for generalists to develop specialized capabilities and insights within a few hours.

The lesson learned from our AI-assisted simulation extends well beyond the retail example above. Every supplier recommendation rests on assumptions. Some matter very little. Others influence nearly the entire recommendation. Traditional sourcing analysis rarely distinguishes between them because its objective is to identify the “best” supplier. Simulation analysis reveals something executives often need more: when the recommendation breaks down and when it holds.

That shift has important implications for how sourcing decisions should be made and communicated.  A sourcing recommendation should no longer end with, “Supplier A is the preferred choice.” It should instead answer four additional questions.

1. Why does Supplier A win?
2. Which assumptions are influencing that recommendation?
3. Under what conditions would Supplier B become the better choice?
4. Which indicators should management monitor after implementation to know whether those conditions are emerging?

These conversations create a different relationship between sourcing and executive leadership. Instead of treating supplier selection as a one-time cost optimization exercise, they recognize it as a strategic decision whose economics shift as markets evolve.

AI can help build models, execute simulations, summarize results, and dramatically reduce the effort required to evaluate thousands of scenarios. It cannot determine which uncertainties deserve attention, which assumptions are realistic, which outcomes matter most to the business, or how much strategic flexibility is worth paying for. Those remain managerial decisions. AI is not replacing that judgment. It is allowing sourcing managers to evaluate decisions in ways that were previously too expensive, too technical, or too time-consuming to be practical.

Executives can accelerate that transition with one simple change. Stop asking your sourcing team which supplier is best. Start asking when and why that answer changes, and what the organization should do when it does. Those questions produce richer analysis, more adaptable sourcing strategies, and ultimately better decisions.

About the authors

Dr. Rodney Thomas is the co-editor-in-chief designate at the Journal of Business Logistics and an associate professor of supply chain management in the University of Arkansas’s Sam M. Walton College of Business. He previously held a number of leadership positions at Lowe’s, IBM, and Michelin.

Dr. Remko Van Hoek is a professor of supply chain management at the University of Arkansas’s Sam M. Walton College of Business and advises companies on procurement transformation. He previously served as a chief procurement officer at a number of companies.

Stephanie P. Thomas (PhD) is an associate professor of practice in the J.B. Hunt Transport Department of Supply Chain Management in the Sam M. Walton College of Business at the University of Arkansas. Her research focuses on challenges related to buyer-supplier relationships in supply chains. She can be reached at Sthomas2@walton.uark.edu.

Today is Thursday, Nov. 5.

1. Record imports expose the supply chain weight of the AI investment boom

U.S. imports reached a record $426.6 billion in September as data-center investment lifted computer and accessory purchases, while worsening supplier delays and tariffs pointed to additional cost pressure.

Reuters  |  Read the original article

2. Canadian factories contract as tariffs deepen cost and delivery pressure

Canada’s manufacturing PMI fell to 49.6 in October as production and orders declined, supplier delays intensified and input costs rose at their fastest pace in more than four years.

Inside Logistics  |  Read the original article

3. DHL holds its outlook as holiday demand offsets fuel-cost pressure

DHL reiterated its 2026 targets, expecting peak-season parcel demand to support fourth-quarter results even after roughly 1 billion euros in additional aviation fuel costs through September.

Reuters  |  Read the original article

4. Service-sector growth slows as supplier lead times and prices remain elevated

The October services PMI eased to 54.5 while supplier delivery times lengthened, new orders cooled and the prices-paid index remained high as tariffs and fuel costs strained supply chains.

Transport Topics / Bloomberg News  |  Read the original article

5. Class 8 orders slide as fleets delay capital decisions

Preliminary October Class 8 truck orders fell 36% year over year to 15,600 units, nearly half the 10-year monthly average, as emissions-rule uncertainty and operating costs constrained purchases.

FreightWaves  |  Read the original article

6. AI-driven workforce cuts raise questions about the supply chain talent pipeline

A Gartner survey found 47% of supply chains reduced staffing by an average 10.5% over two years, with entry-level roles absorbing most cuts as organizations integrated AI into workflows.

Inside Logistics  |  Read the original article

7. Parcel shippers diversify carriers to control unpredictable surcharges

Tractor Supply is spreading parcel volume among multiple carriers so it can redirect bulky shipments when accessorial charges spike, illustrating how network optionality can protect last-mile economics during peak season.

FreightWaves  |  Read the original article

8. European freight softness tempers Trimble’s growth outlook

Trimble cut expected annual recurring revenue growth to about 10% as European freight volumes and spot pricing weakened, even while adoption of AI-enabled procurement and quotation tools continued rising.

FreightWaves  |  Read the original article

9. Calgary airport applies computer vision to cargo turnaround management

YYC will deploy AI-enabled cameras on a dedicated cargo apron to track unloading, loading and fueling milestones, giving operators real-time visibility into freighter turnarounds and potential efficiency gaps.

Inside Logistics  |  Read the original article

10. New resilience service maps multimodal alternatives when trade corridors fail

Lloyd’s Register introduced a planning service that evaluates alternate ports, airports, roads and rail links by capacity, cost, delay and risk, with response plans for a disruption's first 90 days.

Splash247  |  Read the original article

Executive takeaways

Within a few years, artificial intelligence will increasingly handle sourcing, supplier selection, contracting, risk monitoring and many other supply chain decisions. Gartner predicts that by 2030, AI will orchestrate procurement in 30% of organizations (1). 

The business conversation has focused heavily on AI’s ability to improve productivity and reduce costs. Those benefits are real. But they are not the most important question leaders face.

The bigger question is what decision logic organizations are teaching AI to scale.

If supply chains are designed primarily around cost and speed, AI will automate cost and speed. If they incorporate resilience, sustainability, supplier health and long-term risk, AI will scale those priorities instead.

The distinction matters because the forces reshaping supply chains—geopolitical uncertainty, climate disruption, resource constraints and regulatory pressure—are no longer temporary disruptions. They are becoming permanent features of the operating environment.

The evidence increasingly shows that organizations integrating resilience and sustainability into decision-making are creating competitive advantage, not sacrificing performance.

Market has moved

The clearest signal comes from the market itself.

According to the EcoVadis Accenture Sustainable Procurement Barometer 2026, more than 55% of organizations now use AI operationally, and companies with mature AI-enabled procurement capabilities achieve approximately 2.5 times higher revenue growth and 2.4 times greater productivity than their peers.

At the same time, supply chain leaders are operating in a far more volatile environment than they were a few years ago. Trade tensions, tariffs, climate events and supplier disruptions can quickly turn yesterday’s optimal sourcing decision into today’s business risk.

As a result, organizations increasingly need visibility into emissions exposure, resource dependency, labor conditions and geopolitical vulnerabilities, not because stakeholders demand it, but because business performance depends on it.

Yet only 15% of organizations consider themselves advanced at integrating sustainability into procurement decisions.

That gap is significant. It suggests many organizations now possess the data but have not fully embedded it into the choices that shape their supply chains.

Business case is measurable

One of the most persistent misconceptions is that resilience and sustainability come at the expense of growth and efficiency.

The evidence points in the opposite direction.

The Sustainable Procurement Barometer found that organizations proactively building supply chain resilience are achieving 3.6% higher revenue growth than less resilient peers.

At the same time, AI is creating new opportunities to unlock value. Accenture’s experience shows agentic AI can generate 30% to 50% productivity improvements in procurement while helping organizations optimize more than $60 million for every $1 billion of spend (2). 

One consumer goods company unified more than 20 disparate data sources into a single AI-powered interface for its 1,600 buyers, cutting the time each buyer spent on competitive intelligence from more than 10 hours a week to roughly 30 minutes, and unlocking more than $120 million in competitive buying value. AI adoption is accelerating. More than 55% of organizations now deploy AI operationally, with 13% reporting widespread deployment across supply chain workflows.

Meanwhile, Accenture’s Destination Net Zero 2025 research found that nearly 89% of leading companies now explicitly connect decarbonization efforts to business value.

The implication is clear: sustainability is increasingly being treated as a business capability rather than a reporting exercise. Or put differently, sustainability data is becoming resilience data.

Leading companies are already proving it

The strongest evidence comes from companies that have translated these principles into results.

Author of “Net Positive: How Courageous Companies Thrive by Giving More Than They Take”, and former CEO of a large global consumer goods company, Paul Polman described how only about 10% of the company’s products were sustainably sourced when he assumed leadership back in 2009. The company responded with significant investments across its supply chain, stronger supplier standards and unprecedented transparency.

The outcome was not weaker performance. According to Polman, purpose-led brands grew faster, delivered higher profitability and generated approximately $15 to $18 of value for every dollar invested in sustainable production, while shareholders received roughly 300% total return during his tenure with the consumer goods major.

AstraZeneca offers a more contemporary example (3).

The company has embedded sustainable procurement as a core business lever, activating more than 3,500 suppliers across roughly 95% of managed spend. Over 80% of that spend is with suppliers committed to science-based targets, while focused interventions with top-emitting suppliers are accelerating decarbonization where it matters most. The program has delivered a 52% reduction in emissions intensity since 2019, alongside improved efficiency through supplier consolidation and reduced reporting burden.

Both organizations approached sustainability as a mechanism to strengthen supply chains, improve visibility, reduce risk and create long-term value.

The leadership imperative

The debate about whether AI will transform supply chains is effectively over. The more consequential question is whether leaders will use AI to replicate the assumptions of the past or build operating models fit for a far more volatile, constantly changing future.

Technology does not decide what matters next, leaders do and use technology as a tool to get there. The systems being designed today will determine how future procurement decisions balance cost, resilience, risk and long-term value creation. Once those priorities are embedded into AI-enabled workflows, they will be repeated consistently, globally and at a scale that act as timely and measured response to the future disruptions.

That is why this is a leadership decision and needs to be taken today. The leaders who define the next decade will not be those who deploy AI fastest. They will be those who decide, deliberately and transparently, which trade-offs AI should optimize before those choices become automated.

Today is Wednesday, Nov. 4.

1. Labor unrest threatens Bangladesh’s peak apparel production

Wage protests have forced hundreds of Bangladeshi garment factories to suspend production during a critical season, raising shipment risks as manufacturers face higher energy, freight and labor costs alongside buyer pressure.

Reuters  |  Read the original article

2. 3M accelerates manufacturing and distribution network consolidation

3M has closed 16 factories and distribution centers this fiscal year and plans changes at 10 more facilities as it pursues $1 billion in annual productivity gains by the end of 2027.

Supply Chain Dive  |  Read the original article

3. Holiday inventory buildup pushes logistics costs higher

The Logistics Managers’ Index rose to 70.7 in October as inventories expanded before peak season, while contracting warehouse and transportation capacity helped drive aggregate logistics costs to a four-year high.

DC Velocity  |  Read the original article

4. U.S. trade-benefit withdrawal reshapes South African export economics

South African citrus, wine and automotive exporters are assessing margin and employment risks after Washington said duty-free AGOA treatment will end in 2027, although most bilateral exports already fall outside the program.

Reuters  |  Read the original article

5. Carrier-registration lawsuit puts state safety funding at risk

A trucking coalition is challenging the 2027 Unified Carrier Registration fee rule, seeking to halt collections in a case that administrators warn could disrupt registration operations and state commercial-vehicle safety programs.

FreightWaves  |  Read the original article

6. Warehouse vacancy rises without delivering lower operating costs

Flexe’s spot-warehousing report finds higher vacancy has not reduced storage costs, while tariff exposure and changing trade flows are shifting demand away from some gateway markets toward inland logistics hubs.

Logistics Management  |  Read the original article

7. Aircraft demand lifts U.S. factory orders above expectations

September factory orders increased 1.8%, exceeding forecasts, as civilian aircraft bookings surged and core capital-goods orders and shipments advanced, indicating continued business-equipment investment despite broader cost and demand concerns.

Reuters  |  Read the original article

8. AI adoption outpaces the data foundations needed to scale it

A GS1 US survey found nearly 60% of organizations report advanced automation or AI use, but only 38% have centralized real-time data integration and 41% struggle to scale technology pilots.

Supply Chain 24/7  |  Read the original article

9. Warehouse operators improve inventory turns while slowing capital spending

A new warehouse survey found average inventory turns increased to 7.2 while projected capital spending fell sharply, with disruption risk dominating concerns and shipping docks emerging as a frequent bottleneck.

Supply Chain 247 |   Read the original article

10. GXO growth signals sustained demand for outsourced logistics

GXO’s third-quarter revenue rose 5.7% to $3.59 billion as new business wins increased 34% and its sales pipeline reached $3.1 billion, even as operating profit and cash flow declined.

FreightWaves  |  Read the original article

Too much inventory ties up working capital and erodes margins. Too little creates service failures and lost revenue. And with ongoing demand volatility, supply uncertainty, and shifting customer expectations, traditional approaches to inventory management are no longer sufficient.

The challenge is no longer just forecasting demand it’s understanding how inventory decisions impact the entire supply chain, from cost-to-serve and service levels to resilience and responsiveness.

In this session, we’ll explore how leading organizations are rethinking inventory as a strategic lever rather than a static buffer. From improving visibility across multi-echelon networks to leveraging advanced analytics and AI-driven planning tools, companies are building more intelligent, adaptive inventory strategies.

We’ll also examine common pitfalls from over-reliance on historical data to disconnected planning processes and discuss how organizations are aligning inventory decisions with broader business objectives, including financial performance and customer experience.

FEATURING:

Kira Bilecky, Director, Supply Chain Strategy, St. Onge; Tatyana Ventura, Director of Customer Success, RFgen; Kim Hadlum, Vice President of Inventory Planning, Armada and Adrian Wood, DELMIA Strategic Business Development Director, Dassault Systèmes

The business case for warehouse automation needs to account for what happens after installation—including who takes responsibility when performance falls short.

As our sister publication Supply Chain 247 reports, one in four respondents to an Exotec survey wished they had made a different warehouse automation decision. The survey included 250 supply chain decision-makers in the United States and Canada.

For executives evaluating investments, the findings raise two questions worth resolving before signing a contract: Does the business case account for the system’s lifetime costs, and is responsibility for resolving problems clearly assigned?

The research comes from a warehouse robotics provider, so its recommendations should be considered in that context. Still, the questions offer a useful starting point for examining an automation proposal.

Inside the full Supply Chain 247 article

• How ownership and maintenance costs compared with buyers’ expectations.
• Where responsibility for fixing system problems became unclear.
• What respondents reported about warehouse readiness before investing.

Read the full article at Supply Chain 247: Report: One in Four Leaders Regret Their Warehouse Automation Choice

Executive takeaways

Editor’s note: This is part 1 of a 4-part series on multi-tier supply chain networks. Part 2 in the series will publish on Tuesday, Nov. 10, with parts 3 and 4 publishing on Nov. 17 and Nov. 24

Here is a thought for supply chain directors (SCDs). What if we went looking for the bottleneck (BN) in the supply chain network (SCN) now, in calm times, before a supply stoppage ever hits—and began strengthening it?

Brian (Yoshiatsu) Kawabata

If you have spent time in this industry, you may know the state that Boeing and others call recovery: the emergency footing a stretch of the network drops into once backlog has piled up at a supplier and a supply stoppage looms. In recovery, the original equipment manufacturer (OEM) sends its own engineers into a struggling supplier to raise output and claw the schedule back. Boeing has run a standing “supplier recovery team” for exactly this. It works—but it is, by definition, a response after the damage is done. If you have lived through the chaos of one even once, wouldn’t you rather act one step before it begins?

The risk you can control

Let me set one thing aside first. This is not about disruption from disaster or accident. That kind of risk is real, but it falls across every firm more or less randomly, and at roughly equal probability. You can insure against it, site a plant away from known hazards, do what is reasonable—yet your rivals face the same exposure, and there is no end to that kind of preparation against the uncontrollable risks.

What is inexcusable, for an SCD, is leaving unaddressed the risk you can control. Picture a marching band of players of very different fitness, gradually asked to march faster. Sooner or later the least fit member falls behind. Now imagine lightening that member’s load, or giving them a boost to keep pace, before the parade ever sets off.

Planning a ramp-up is exactly that moment. As you raise the production rate across the network, the weakest process—the node with the least headroom—will, at some point, fail to keep up and begin to delay. This is controllable for two reasons: the ramp is planned, so you know it is coming; and the weak point, if you go looking in advance, may well be found. If you know today where it is, you can concentrate your investment there. Diagnosis precedes treatment; making the slowest node visible, before it bites, is that diagnosis. It may be the single most important discipline in the next generation of supply chain management.

Consider why this matters competitively. If a bear is chasing you, run—and for an SCD, running is not optional but a duty. The person beside you may laugh: the bear is far faster than any human, so why bother? But I do not need to outrun the bear. I only need to be one step ahead of the person beside me. Handling your controllable risks will not prevent every disruption. But when a risk you managed well is one a rival stumbled on—when they could not ramp and left their customers waiting—you take their share, and improve your own odds of survival.

When a network gets too deep to see

I have worked on supply chains from both sides of the buyer-supplier relationship, and the need for visibility was opposite at each. At Intel, the upstream network is relatively small; the people responsible knew their suppliers’ situations—and those of the equipment makers their suppliers bought from—in considerable detail. The network was shallow and narrow, and there was little role for a network-wide tool to locate a BN. At Boeing, it was the reverse. The number of suppliers behind a single aircraft is of another order entirely. In the automotive industry the count can run to thousands of firms. And beyond Tier 1, visibility grows dim; one cannot claim to see clearly up there. The breadth of the search space governs how deeply one can see into it. Visibility, in other words, is not universally necessary. It becomes necessary only once a network passes a certain depth and breadth—and today’s complex products, cars and aircraft, are long past that threshold.

The hundred-page survey

Take the Rate Readiness report. At the aircraft manufacturer, we, the OEM, imposed it on our Tier 1 heavy-industry suppliers: a survey well over a hundred pages, asking how ready they were for a future rate increase, which they had to answer. Each department filled in its dozens of pages; sales pulled it together; a single round took three months. It covered capital investment and much else—a heavy burden for the supplier. And if that worked, why does supply disruption still surface the way it does? For one, the survey may not carry all the way through to the Tier 2 suppliers above Tier 1 and beyond. And by demanding that every item be answered, by every Tier 1 firm, does it not lower the quality of the whole? When you learn a language, which teacher would you rather have—the one who says to memorize the whole dictionary, or the one who narrows it to the grammar that matters? Rather than burden every firm with an exhaustive report, shouldn’t we narrow the burden to just what locating the constraint requires, and lighten the load?

Why suppliers keep the door closed

I have visited many suppliers. Ordinarily, most were reluctant to allow us access or to share information. Propose that we find the BN process together and strengthen it now, in calm times before recoveries, and getting agreement was extremely hard. Almost no firm ever took us up on it. Only in a crisis—that recovery state, backlog piling up, a stoppage looming—would they grudgingly allow a visit and let us measure process capacity. One supplier, after we had grown close over a long time, let me in on what suppliers rarely say out loud: “You can tell me to hand over the internal process information. But at next year’s contract renewal you’ll use it to negotiate the price down, won’t you? So how could I hand it over easily?” That is entirely reasonable. A one-sided demand for disclosure is nothing but exploitation. And there is always the risk the price leaks to one of the supplier’s rivals, who undercuts them and takes the business.

Read Part 2: Projecting constraints in multi-tier supply chain networks: Two traditions, one measurement

So an OEM simply demanding the information from a supplier cannot work; it is structurally impossible. Even if you asked every firm in the SCN to reveal its internal information, each would push back: why come to us? Do we even have to answer? Ask all of them equally, and one firm reckons it can stay quiet without standing out, while another braces for the reverse—that having answered honestly, it alone will be the one to lose. To get cooperation, you need one of three things: a special case with overwhelming leverage to compel it, some benefit to the firm from answering, or grounds strong enough to justify the request.

By now the problem is clear. If you could locate the BN across the whole network in calm times, you could act before the ramp. The value of visibility is plain. But when the OEM asks a supplier to disclose its internal information, no answer comes back—not out of obstinacy, but because disclosure is structurally a loss. So: rather than forcing every firm to open everything, might there be a way to find the BN while asking for only a little?

What I want to propose over this series is an information-sharing platform for exactly that. Rather than ask every firm to open its interior, it takes a single number from each, and from those alone draws out where the constraint of the whole network sits, in peacetime. Everyone comes to know who carries the weight of the whole—and no firm loses by disclosing.

As it happens, the key has been sitting in plain sight for more than half a century. It lies in two old tools that grew up separately, in project management and in production management, and have rarely been brought together. Next time, I want to propose that these two are in fact saying the same thing. Once that is clear, a single number from each firm may be enough to draw the BN candidates across the entire network.

About the author

Brian (Yoshiatsu) Kawabata began his career at Honda, earned an MBA at Toulouse Business School in 2008, and went on to work at Boeing and Intel. In June 2026, he completed a PhD in computer science at Tokyo University of Agriculture and Technology, with a dissertation on detecting the constraints of multi-tier supply chain networks from minimal shared information; a related paper appears in Information Technology and Management (Springer, 2026; https://link.springer.com/article/10.1007/s10799-025-00478-x). He writes from both sides of the work—the factory shop floor and the theory.

Today is Tuesday, Nov. 3.

1. U.S. manufacturing extends its expansion despite persistent cost pressure

The October manufacturing PMI rose one point to 55.5, marking a 10th straight month of growth, while strong production and new orders contrasted with elevated input prices and continued tariff uncertainty.

Logistics Management  |   Read the original article

2. Asian manufacturers’ U.S. investment pipeline continues to build

Citi expects Japanese, South Korean and Taiwanese companies to keep increasing U.S. investment as factories move from land purchases into construction and operations, sustaining demand for trade finance and cross-border services.

Reuters  |  Read the original article

3. Port labor talks avert an immediate Southern California disruption

Employers and union watchmen extended their contract through Nov. 4 while federally mediated talks continue, temporarily reducing strike risk at terminals handling large shares of Los Angeles and Long Beach cargo.

FreightWaves  |   Read the original article

4. Truckload capacity exits shift negotiating leverage toward carriers

About 89,000 carriers have left the market since 2022, creating a supply-led recovery in which rates are firming despite uneven freight demand and increasingly regional differences in capacity availability.

Logistics Management  |  Read the original article

5. Reshoring and foreign investment sustain the manufacturing job pipeline

Manufacturers and foreign investors announced 247,000 U.S. jobs in 2025, with transportation equipment leading and the Reshoring Initiative projecting 25% growth in related announcements during 2026.

Supply Chain Dive  |   Read the original article

6. Highway funding lapse raises risk for freight infrastructure projects

Congress returns facing a lapse affecting $10.8 billion in nationwide formula funding for roads, bridges and safety, while another short-term extension could leave freight corridors and truck-parking projects uncertain.

Transport Topics  |   Read the original article

7. Holiday parcel demand rises as carrier market shares diverge

ShipMatrix forecasts 2.4 billion peak-season packages, up 4% year over year, as Amazon and alternative networks gain volume while retailers absorb sharply higher parcel rates and fuel-related charges.

FreightWaves  |   Read the original article

8. Foreign investment agreements target U.S. production and critical supply chains

Mitsubishi plans a $490 million Georgia expansion, while U.S. agreements with Finland and Denmark seek stronger critical-minerals, technology and shipbuilding capacity alongside Hanwha’s new naval construction contract.

Manufacturing Dive  |   Read the original article

9. AI moves digital freight matching deeper into brokerage workflows

Industry analysts say automated matching, pricing and carrier vetting are becoming embedded in brokerage operations, while tighter capacity and liability concerns increase the value of dense networks and human exception management.

Logistics Management  |   Read the original article

10. MSHA urges winter controls for mine operations and equipment

A federal safety alert warns that snow, ice and freeze-thaw cycles increase vehicle, ground-control and cold-stress hazards, recommending equipment winterization, surface treatment, cautious driving, inspections and worker training.

Work Safety 24/7  |  Read the original article

For a deeper dive into this topic, read: Continuity plans are not enough: Bringing supply chain into the enterprise risk conversation

 

My Ph.D. from Wharton is in operations research, which I received a long time ago. My graduate studies included learning about constraint-based optimization algorithms (such as linear programming) and decision analysis theory. There were a few of the more theoretical learnings that I had not seen widely used during my working career.

However, during retirement, I watched the COVID-19 pandemic unfold and noted that traditional business planning processes—such as demand forecasting and the sales and operations planning (S&OP) process—were rendered largely useless. Companies struggled to perform supply demand planning for uncertain futures. They also struggled with severe supply shortages (such as semiconductor chips) and lockdowns that caused labor shortages. Prior S&OP planning assumed supply was plentiful, and that forecasting could be done using historical demand. Thus, I realized that at least two special planning teams would have to be assembled to support forecasting and planning under uncertainty, and during severe supply shortages.    

Two special planning teams proposed

In the May/June 2024 Insights column, “Special planning teams for unusual times,” I proposed adding two special ad hoc planning teams to the “hierarchy demand-supply planning framework” presented in it: “Quick Response under uncertainty” and “Supply-in-mind demand optimization under risk.” This approach would keep intact the traditional roles of the S&OP process as the lynchpin tactical-planning process that ties corporate strategic plans to execution-based operational plans. Also, it would not disrupt the activities of an S&OP team, which is not chartered nor experienced in handling uncertainties and severe supply constraints.

Rationale for not disrupting S&OP

S&OP is an important rudder for navigating a company toward financial objectives. The team develops detailed planning numbers that drive operations over a 12 month to 18 month planning horizon. Expecting it to incorporate short-term uncertainties and supply shortages would severely hamper its job of updating detailed plans. S&OP is a tactical planning process under risk, and not under uncertainty. Decision-making under risk is predicated on being able to develop random probability distributions from historical data. Under uncertainty there is negligible useful historical data from which to draw.

In addition, the S&OP process was largely predicated on supply always being available to meet unconstrained demand. Both the short-term “demand control” and SOE operational planning processes also assume decision-making under risk, and that supply is plentiful.

Because a strategic planning process has a planning horizon of three or more years, forecasts are not useful, and uncertainties abound. Thus, strategic plans are traditionally developed using a scenario-planning approach and often need a special project team that might require external third-party consulting help as well. 

The Quick Response planning process

In my Insights column, “Under uncertainties: Quick Response, not only S&OP” (March/April 2022), I argued that the S&OP process should not be used for planning under uncertainty.

I recommended a Quick Response team be assembled with responsibility for operational planning during extreme uncertainties in demand and supply. I termed it Quick Response because managers must make fast decisions with little to no historical information. Decisions made are usually practical and involve common sense. For example, during the early days of the COVID-19 pandemic, policymakers could only advise the public to stay away from each other, wear masks, wash their hands, and don’t go to work. Later, policymakers had sufficient information to advise getting vaccinations, as well.

A Quick Response process would be temporarily put in place whenever uncertainties arise in a significant portion of a demand-supply chain. An executive-led team of specialists would be assembled to do short-term planning. It would exist until enough information and data had been assembled to return the planning back to the S&OP team.

To demonstrate how scenario planning might be used in planning decisions under uncertainty for the pandemic, I wrote an Insights column titled: “Decision making under uncertainty: A primer” (May/June 2022). I discussed a hypothetical $1 billion profit-sized company, looking at three types of strategies to deploy throughout the pandemic.

The supply-in-mind demand optimization process

In my March/April 2024 Insights column, “Supply constraints? Demand-shaping revisited,” I discussed another special team that might be required from time to time. It was needed during the COVID-19 pandemic to do short-term operational planning for products with severe supply shortages. S&OP teams are typically not experienced with supply constrained demand matching. Rather they are used to matching future supply with demand forecasts, and then buffering inventory, capacity, and lead times to ensure demand is met.

Generally, a supply-in-mind demand-shaping process involves identifying supply issues and developing opportunities to enhance demand. By aligning supply with marketing & sales activities. It might, for example, be aimed at optimizing demand-side objectives such as revenues, operating margins, and market share.

Additionally, when there is an excess of materials and components, underutilized plant capacities, and a surplus of finished goods inventory, supply managers might collaborate with sales & marketing managers to develop and promote programs aimed at taking advantage of these excess supply opportunities.

However, if there is a shortage of any type of supply, marketing & sales ought to change demand plans to less aggressively promote and sell any products affected because if demand exceeds supply, managers will have to take emergency actions to meet any excess demand, and these will be more costly and less profitable.

Constrained-based optimization needed

During supply shortages, a critical starting question for planners to address would be: “What product mix can be made from readily available finished goods, as well as supplies on hand, on order, and procurable in the future?” A variety of product mixes would be identified to help S&OP teams assess and select which is best. To do this, material requirements planning (MRP) and bills-of-material systems might have to be run in reverse. However, I suspect that most MRP systems might not be easily modified to work in reverse. Nor would it be cost-effective to develop a new system solely dedicated to operational planning purposes. I recommend a more practical approach that would involve a class of well-known constrained-based optimization algorithms. The most used one is the Simplex Method that is used to solve linear programming (LP) problems, such as Microsoft Excel’s free add-in Solver.

LP problems involve only linear equations of decision variables (“Quantitative Analysis for Management 13th Edition” by Barry Render, Ralph M. Stair, Jr., Michael E. Hanna, and Trevor S. Hale. Pearson, 2017). One is the objective function to be optimized, such as maximizing revenue and profit, or minimizing costs and assets. Others involve a set of constraint equations that represent limits upon which the optimal solution is feasible. For example, one equation for selecting an optimal investment portfolio might be as simple as the maximum investment in risky gold stocks, or the minimum investment in safe government bonds.

One of the most used LP formulations is the “product mix problem.” It seeks to determine what product mix will maximize profits subject to various supply and marketing & sales constraints. This LP approach could be used by a demand optimization (DO) special team to develop alternative product mixes that can be made from available supply.

A car manufacturer LP illustration

During the early days of the COVID-19 pandemic, car manufacturers experienced serious semiconductor chip shortages. In addition, as the pandemic ensued, labor shortages arose on assembly lines from lockdowns. As an illustration, a fictitious car manufacturer’s DO planning team has determined that five of its car types were being affected by chip and labor shortages: full-sized sedans; compact sedans; SUVs; electric vehicles; and hybrids.

It wants to provide the S&OP team with how many of these cars to make next month. The DO team decided that the company should make at least a minimal mix of 100,000 cars next month, and then optimize “car operating margins” based on the chips and labor that would be available.    

 

Information on the various parameters including operating margins, minimum production levels, semiconductor chips per car as well as the labor hours needed to make a car are shown in Figure 1. Additionally, the availability of chips for next month’s production is 150 graphics chips and 150 proprietary chips. Also, only 750 hours of unskilled labor and 200 hours of computer tech labor will be available next month. 

The LHS of the constraint equations represents the number of cars manufactured and the chip supply utilized. The RHS of the constraint equations represents the minimal car production levels and the available chip supply. The optimal cars to make represent the ideal quantity of vehicles to produce while the optimal car operating margin shows the operating margin for each vehicle type, resulting in a total car operating margin of $2,370. (All numbers are in thousands.)

The DO team also wanted to provide advice to the S&OP team about additional supply that might be added to enhance operating margins, for example, via additional overtime labor hours or spot-market chip purchasing. To do this it looked at the LP spreadsheet and the sensitivity report that “solver” created.

In comparing the LHS to the corresponding RHS, these showed that chips available were not all used in the optimal production mix. This meant that chip supply was not affecting car margin, so increasing the supply of these would not enhance margins. According to the sensitivity report, each hour of computer tech labor would yield $12,000 more in operating margin—minus extra overtime costs. This means that the S&OP team ought to assess whether to add more computer tech overtime hours. The team would need to collaborate with marketing & sales about promoting and selling the optimal mix, as well as collaborate with manufacturing to get a commitment to more hours of overtime by its computer tech workers.

 

Portending a future with more DO teams

In my November 2023 Insights column, “Global economies also include longevity,” I discussed that the world is getting older because of declining birth rates and people living longer. This is especially true in industrialized economies—wherein significant portions of their adult populations will be retired and significantly smaller portions of young people are working in supply chains. Thus, future economies might find it difficult to sustain an innovative growing workforce to meet the growing longevity demand. While there was a need for DO teams during the pandemic, I suspect few if any companies had them. However, I suspect that these special teams will be required in the future—to plan for the impending labor and supply shortages.

Executive takeaways

The Technology and Telecommunications industries have entered a new era of supply chain strategy. For decades, companies have built supply networks around efficiency, scale, specialization and access to concentrated manufacturing ecosystems, which delivered extraordinary cost and innovation advantages, but it is now being tested by a growing complexity.

Global export restrictions, tariff escalation, critical-mineral concentration, fuel-price volatility and the fragmentation of global trade networks are structural features of this more complex operating environment. Supply chains are now being tested to absorb a level of uncertainty they were never built to handle. (1)

Making consequential billion-dollar decisions under uncertainty and ambiguity is not new. Arguably, this has always been the job description of supply chain executives. What is new is that the decision window is shrinking, the consequences of those decisions now span the whole enterprise, and the information foundation is not keeping pace. (12) Leaders have become more pessimistic since last year, with 34% now saying their supply chain isn’t ready for the future, up from 27% in 2025. (12)

Similarly, confidence in future readiness and in the ability to respond to disruptions has fallen from 73% to 66% year-over-year. (12)

By the numbers

The shift is already underway. AI-driven innovations have already delivered significant results, including a 5% to 10% reduction in operational costs and over 10% returns on AI investments for 67% of respondents. (8) Eighty-nine percent of technology and telecom companies are launching digital capabilities, including AI agents, data ecosystems, and intelligent automation. Of these, 29% are extending these capabilities to suppliers and partners to improve planning, allocation, and coordination across constrained supply networks. (9)

Why it matters

Technology and Telecom supply chains rely on diverse raw materials, complex components, extensive contract manufacturing, and a deeply interdependent tiered supplier base. Yet many organizations still manage these networks with fragmented data, traditional sourcing playbooks, and visibility that stops too early. Supply chain leaders who lack confidence in their network operations and subsequently their own decision-making cited poor end-to-end visibility as a root cause. (12) In fact, leaders who expressed growing confidence in their supply chains and decision-making also cited their initiatives and improvements in end-to-end visibility. (12)

Critical minerals illustrate the problem. The geopolitical controls imposed in December 2024 on gallium, germanium, antimony, superhard materials, and dual-use graphite items showed how critical raw materials can become leverage points for technology and semiconductor supply chains, and traditional sourcing approaches can often fail to react quickly to the supply chain disruptions. (2) Deloitte research found that only 15% of procurement leaders had visibility beyond Tier 1 suppliers (3), while Gartner’s 2025 supply chain survey found that deeper-tier visibility remains a persistent weakness even as Tier 1 transparency improves. (4, 6)

The result is a decision-making paradox

Among the strategic priorities for supply chain leaders, the trend of a sharp year-over-year shift in the priority of “decision making speed” is notable. For supply chain leaders, this has jumped five spots, becoming the second biggest priority in 2026, just behind “improving efficiency and profitability.” (12) Investors expect leaders to move faster, commit larger amounts of capital, protect margins, and guarantee continuity, but they often lack the integrated intelligence needed to see the full consequence of their choices. A sourcing decision is also a geopolitical decision. A product launch decision is also a capacity-allocation decision. A supplier switch may affect compliance, carbon exposure, warranty risk, and customer commitments.

Simply, disruptions occur at a faster rate and at the same time complexity is increasing. As such, consequential decisions are needed faster and often with less-than-preferred visibility into the long-term consequences.

The AI opportunity

Enterprise AI adoption varies widely, with significant gaps in deployment. Successful enterprises start with cohesive data strategies, often implementing semantic ontologies or knowledge graphs to create relation between data in which ML and predictive AI can leverage.

Sitting on top, agentic AI can then autonomously flag and action issue resolution. This is where AI can change the nature of supply chain decision-making. The next frontier of visibility is not simply “knowing where things are.” It is understanding why a risk is emerging, how it could propagate, which business outcomes it threatens, and what action should be taken before disruption reaches the enterprise. Some companies have been successful by building the foundation, pairing data with decision governance, and enabling AI Agents so the benefits of agentic AI can be fully realized.

HPE has demonstrated a successful deployment of AI in its supply chains, showing how to reduce complexity by shifting from information production to decision orchestration and, in bounded circumstances, decision execution. HPE positions GreenLake Intelligence as a step beyond traditional AIOps. Rather than just monitoring and automating IT operations through machine learning, its agents are designed to act autonomously, understand context, and make decisions proactively. (5) HPE Operations uses AI in 50 workflows to improve quality, speed, and efficiency. Key cases include support incident detection and AI-driven manufacturing with deep learning and computer vision. A “human-in-the-loop” approach can help ensure accuracy and optimal results. (10)

Creating a foundation for agentic AI to succeed is also driving significant value to the enterprise. Cohesive data architecture, semantic relational data, and clear decision-making processes with transparent accountability are playing a huge role in supply chain preparedness. For instance, take the case of customer commitments, where machine learning (ML) models trained on allocation shifts, order changes, and historical supplier variability can flag which promised delivery dates are most likely to slip before the delay becomes visible. That early signal lets teams intervene with supply partners and set realistic expectations before disruption reaches the order book. (11)

Another use case is around AI-enabled planning assistants that can summarize plan changes in plain language, surface likely root causes in minutes, and give planners more time to act with greater confidence in the plan. Beyond predictive analytics and planning efficiencies, AI tools can help companies respond faster to emerging crises, increasing their supply chain resiliency. For instance, AI tools enable companies to assess and compare response strategies through simulations. These simulations estimate how each strategy may affect supply, demand, supply chain partners, and recovery times following a disruption. By evaluating the results, firms can quickly develop a well-informed response to sudden changes. (7)

The value of AI is not that it replaces executive judgment. It can improve the quality, speed, and consistency of that judgment. This is the transition from visibility to reasoning.

What's different in 2026

Three years ago, “AI in supply chain” mostly meant better dashboards and earlier alerts, but still a human reading a screen and deciding what to do. Gartner’s 2026 supply chain technology trends flag a real inflection. AI is moving from generating insight to taking a role in operational execution itself. (13) Gartner projects that by 2030, half of cross-functional supply chain management solutions will use intelligent agents to autonomously execute decisions, not just recommend them. (14) So the real 2026 story is not whether AI can see a problem, but how much decision authority a company is willing to hand it, and what has to be true (i.e., clean data, clear guardrails, an audit trail) before that hand-off is safe.

In practice, most organizations pursuing this today are settling on a three-tier governance model rather than an all-or-nothing switch to autonomy. Routine and reversible decisions are ones where the agent acts and logs the action. Time-sensitive but moderate-risk decisions are ones where the agent acts unless a human intervenes within a set window and high-stakes or novel decision are ones where the agent prepares the analysis but a person still decides. (15) This tiering, matching the level of autonomy to the shape of the decision rather than the ambition of the technology is what separates real agentic deployments from what Gartner has started calling “agent washing,” rebranding ordinary automation or chatbots as autonomous agents without any actual decision authority changing hands. (17)

Where trends are going to

Given the increasing complexity and consequences of decision-making, supply chain leaders must approach solutions from a foundational perspective, i.e., build the foundation so that the benefits of agentic AI can be fully realized. Often, enterprises can fall victim to a general sense of urgency or falling behind. In this case, it’s important to take stock and move methodically.

• Start narrow, then expand. One contained pilot (late-delivery prediction, exception handling, inventory replenishment) can go live in three to six months. Several connected workflows take six to 18 months while full orchestration may take 18 to 36 months. (15) The goal isn’t full autonomy on day one, it’s proving the AI is reliable at the authority it’s given before extending it. It’s worth noting that Gartner expects over 40% of agentic AI projects to be abandoned by 2027, usually from unclear business cases or weak guardrails. (16)

• Pair cohesive data with decision governance. Placing AI on top of poor data architecture will only exacerbate flawed decision-making. As data improves, formal decision frameworks are needed to keep things auditable and accountable.
• Enhance AI agents. Introduce AI agents to handle data reconciliation and routine decisions, effectively cloning senior planners’ judgment so scarce expertise scales further.
• Evolve control systems into decision engines rather than just flagging a disruption. Engineer existing systems to simulate response options (reroute, reallocate inventory, prioritize orders) that can act automatically.
• Connect dots. Unified/connected planning of AI-driven impact simulation makes tradeoffs between cost, service, risk, and sustainability explicit before a decision executes, rather than discovered after.

Volatility is likely to remain a structural feature of this sector. The winners will likely be companies that can sense change earlier, understand exposure more deeply, simulate options faster, and act with confidence before disruption becomes damage.

For supply chain leaders, AI can help teams detect supplier concentration risk before it becomes a shortage. It can help operations leaders model the impact of geopolitical and logistics disruption before committing to production plans. It can help finance leaders understand the working-capital impact of resilience strategies. And it can help boards move supply chain risk from periodic review to continuous strategic oversight.

The next frontier of supply chain leadership will not be visibility alone. It may be the ability to reason, decide, and act with confidence in an environment where disruption has become constant.

About Deloitte

Deloitte refers to one or more of Deloitte Touche Tohmatsu Limited, a UK private company limited by guarantee (“DTTL”), its network of member firms, and their related entities. DTTL and each of its member firms are legally separate and independent entities. DTTL (also referred to as “Deloitte Global”) does not provide services to clients. In the United States, Deloitte refers to one or more of the US member firms of DTTL, their related entities that operate using the “Deloitte” name in the United States and their respective affiliates. Certain services may not be available to attest clients under the rules and regulations of public accounting. Please see www.deloitte.com/about to learn more about our global network of member firms.

Author contacts

The following authors can be contacted with any questions regarding the article:

Cody Stock, codystocks@deloitte.com

Francesco Stefanelli, fstefanelli@deloitte.com

Nick Dunn, nicdunn@deloitte.com

References

1. https://www.scmr.com/article/geopolitical-readiness-in-supply-chains-strategicchallenges-for-leaders

2. https://www.sdcexec.com/safety-security/regulations/article/22950678/bloomberggeopolitical-risk-in-supply-chain-management-is-entering-a-new-era-of-human-and-aiintelligence

3. https://www.deloitte.com/us/en/insights/topics/operations/chief-procurement-officercpo-survey.html

4. Gartner Says Supply Chain Leaders Should Prioritize Advanced Data Visibility and Scenario Planning to Drive Competitive Advantage Amid Global Uncertainty

5. GreenLake Intelligence – The agentic AI architecture for autonomous hybrid cloud operations

6. https://cset.georgetown.edu/publication/china-rare-earth-export-ban/

7. https://gjia.georgetown.edu/business-economics/the-role-of-ai-in-developing-resilientsupply-chains/

8. https://economictimes.indiatimes.com/news/company/corporate-trends/over-half-ofindias-technology-media-and-telecommunications-organisations-fully-integratedai/articleshow/114310254.cms?from=mdr

9. https://www.gartner.com/en/newsroom/press-releases/2026-04-07-gartner-forecastssupply-chain-management-software-with-agentic-ai-will-grow-to-53-billion-in-spend-by-2030

10. HPE as a manufacturer uses AI in its own operations | HPE

11. https://www.gartner.com/en/newsroom/press-releases/2026-04-07-gartner-forecastssupply-chain-management-software-with-agentic-ai-will-grow-to-53-billion-in-spend-by-2030

12. 2026 Supply Chain Compass: How Supply Chain leaders are navigating complexity

13. Gartner Identifies Top Supply Chain Technology Trends for 2026

14. Gartner's Top Tech Trends: Autonomous Supply Chains — Gartner estimates 50% of cross-functional supply chain solutions will use intelligent agents to autonomously execute decisions by 2030

15. Agentic AI in Supply Chain Management: three-tier governance model and pilot-to-production timelines (2026)

16. Execution, not chat: How Agentic AI changes supply chain operations — Supply Chain Management Review (40% agentic AI project abandonment rate by 2027, per Gartner)

17. Gartner Warns of Agent Washing Risks in Supply Chain Planning Technology Market

Today is Monday, Nov. 2.

1. U.S. and Canada outline a possible trade agreement

U.S. and Canadian leaders said their teams will pursue a trade agreement after recent discussions, potentially easing bilateral tariff tensions while linking negotiations to energy infrastructure and cross-border investment.

Reuters  |  Read the original article

2. China’s factory activity contracts as export orders weaken

China’s official manufacturing index fell to 49.3 in October from 50.1 in September, while new export orders dropped to 48.7, signaling softer demand across a major global production base.

Associated Press  |  Read the original article

3. USDA funding targets greater domestic fertilizer capacity

USDA proposed $573 million for eight fertilizer projects representing more than $1.3 billion in investment, with the facilities expected to add domestic nutrient capacity and reduce reliance on imported supplies.

Supply Chain Dive  |  Read the original article

4. Holiday logistics hiring shifts closer to peak demand

Retailers and logistics providers are delaying seasonal hiring and using faster recruiting, flexible schedules and automation to adjust labor closer to demand, reducing the risk of overstaffing or understaffing an uncertain peak.

The Wall Street Journal  |  Read the original article

5. Schneider shifts its eastern intermodal strategy to Norfolk Southern

Schneider will make Norfolk Southern its primary eastern rail provider in January 2027, pairing Schneider’s asset-based network with NS capacity as intermodal demand grows and a proposed transcontinental rail merger remains pending.

Logistics Management  |   Read the original article

6. Stricter cabotage enforcement may tighten cross-border truck capacity

Knight-Swift’s chief executive said increased scrutiny of Mexican drivers, including visa actions tied to alleged past cabotage violations, could reduce available cross-border capacity as tariff uncertainty changes shipper strategies

FreightWaves  |  Read the original article

7. UPS expands upstream RFID labeling to improve parcel visibility

UPS says customer-site RFID labeling now covers 3 million packages daily at 315 locations and could reach 6 million per day in 2027, providing earlier pickup scans while reducing manual handling.

Supply Chain Dive  |  Read the original article

8. Georgia inland port opens a rail alternative for manufacturers

Georgia’s $134 million Gainesville Inland Port links regional manufacturers to Savannah by Norfolk Southern rail and is expected to shift 26,000 containers from highways during its first year of operation.

FreightWaves  |  Read the original article

9. AIT acquisition extends project-logistics reach into Oceania

AIT Worldwide acquired Australian forwarder Connor Anderson International, adding air, ocean, road, customs and project-cargo capabilities in Sydney and Melbourne as technology customers expand data-center construction across the region.

FreightWaves  |  Read the original article

10. A five-step framework reframes procurement cost reduction

A contributor proposes that purchasing teams address costs sequentially by eliminating unnecessary requirements, consolidating where risk permits, qualifying alternatives, negotiating from a defensible baseline and managing supplier performance after award.

Supply Chain Management Review  |  Read the original article

Executive takeaways

In a hyper-competitive world, it is becoming increasingly hard to gain new business and even harder to retain existing market share. This is especially true for organizations selling physical products. While the world is hyper-globalized—allowing sellers to access markets across the globe—the reality is that our environment is becoming increasingly VUCA. Coined in the late 1980s by the U.S. Army War College, VUCA stands for Volatile, Uncertain, Complex, and Ambiguous. The events of the last 24 months, exemplified by active geopolitical conflicts, AI-driven disruption, demographic upheaval, and trade wars, encapsulate this phenomenon well.

The emerging consensus among experts is that supply chains and global commerce will come under increasing pressure over the next decade or two. Technology continues to reshape consumer demand, while geopolitics constrains market access and forces the creation of regional production capabilities. However, it is not all gloom and doom. With deliberate preparation and an integrated, nimble approach, purchasing executives can prepare their businesses to thrive in this environment and capture market share as competitors falter.

Refashion cost management from first principles

Thriving in a VUCA world requires supply chain and purchasing leaders to act decisively with vision and foresight. One specific area with a direct impact on both revenue and profitability is robust cost management. Over the decades, several frameworks have emerged—Bottom-Up Costing (BUC), Total Cost of Ownership (TCO), Design to Cost/Value (DTC/V), and Value Analysis/Value Engineering (VAVE), to name a few.

While each of these techniques has immense merit—providing purchasing professionals with a thorough understanding of product costs and effectively isolating cost centers within supply chains—they are often reactive in nature. That is, they only reflect the status quo as it currently exists. For a more forward-thinking approach, supply chain and purchasing executives must question all layers of cost drivers—including the feature offerings of the product, associated manufacturing methods, and supply chain network options—and deliberately construct a cost stack from scratch. Such a first-principles approach can be envisioned as a set of interconnected cost-reduction priority sets, ultimately driving best-in-class operating costs and financial performance.

Cost reduction priority sets

Typical quarter-end organizational cost reviews focus on identifying high-cost line items and attempting to eliminate them in the subsequent months. I advocate for a more structured framework to achieve an optimized cost structure through a sequence of strategic choices called Cost Reduction Priority Sets. In this approach, the practitioner moves sequentially through Priority Sets 1 through 5, making decisions based on the hierarchy: Eliminate → Consolidate → Alternate →Negotiate → Manage through disciplined execution.

 

Eliminate—Priority set #1

The first step in any cost management exercise should be the ruthless identification and elimination of features, products, and processes that are superfluous or offer duplicate value. Eliminating features yields a 1:X payoff. Evidence in technical literature points to a clear non-linear cost multiplier, where every $1 reduced in bill of materials (BOM) cost can result in a 2X to 4X reduction in overall product cost (Boothroyd, G., Dewhurst, P., & Knight, W. (2010). Product Design for Manufacture and Assembly (3rd ed.). CRC Press).

While this phenomenon may seem counterintuitive to the uninitiated, supply chain and operations practitioners frequently observe that feature elimination unlocks a cascading sequence of consequential cost reductions. These include removing secondary supporting features, reducing tooling expenses, saving on yield fallout, lowering inventory holding costs, and decreasing design, manufacturing, and maintenance man-hours.

Furthermore, the cumulative impact of these savings on unit cost reduction and revenue acceleration is substantial. In consumer goods—where the end-consumer price is often 4X the manufacturing cost due to margin stacking across contract manufacturers, brand owners, and distribution channels—every dollar saved in the feature set can result in an 8X to 16X reduction in the price paid by the end consumer. This becomes a powerful accelerator for driving market share and revenue growth. In short, focusing on eliminating non-value features, processes, and testing fuels a virtuous business cycle.

Consolidate—Priority set #2

The second lever for reducing costs—especially early in the design cycle—is driving consolidation. This occurs first through part consolidation, and subsequently through supplier and supply chain consolidation.

Part consolidation requires a thorough examination of part counts, particularly mating or adjacent components that can be merged into a single integrated component. In many ways, part consolidation achieves feature elimination by another mechanism. The benefits flow directly to the balance sheet via lower part counts, reduced engineering hours, and minimized inventory requirements.

An even greater benefit can be realized through supplier consolidation. While purchasing executives often hesitate to concentrate volume under a single supplier due to dual-sourcing policies, the advantages of supplier consolidation are substantial, particularly for younger, leaner companies. Consolidation not only drives down unit costs through volume leverage, but it also unlocks several operational benefits: streamlined material flow, lower freight costs, simplified inventory tracking, and enhanced material traceability in the event of quality excursions. Operations leaders should execute this step immediately after feature elimination to create a compounding cost-reduction effect.

Alternate—Priority set #3

The third step in the framework involves exploring alternative solutions. Alternatives often take the form of overlooked or non-traditional components, suppliers, or manufacturing equipment.

When sourcing high-demand or critical components, teams frequently face scarcity-driven price spikes. Instead of attempting to negotiate marginal cost reductions on tight-supply items, organizations can source more abundantly available, lower-cost “generic” alternatives. While true generics do not exist for every application, many components are produced in industrial, automotive, and consumer-tech variants. These variants are often manufactured by the same supplier using similar production techniques, differing primarily by industry-specific customizations.

A similar logic applies to sourcing suppliers with high idle capacity (and thus lower pricing) outside traditional industry boundaries. For example, when sourcing automotive components, an organization might utilize available capacity at an aerospace supplier, a farm equipment manufacturer, or an industrial machinery fabricator. Layering Priority Set #3 on top of Sets 1 and 2 further amplifies total cost-reduction potential.

Negotiate—Priority set #4 

The fourth step in the journey is engaging in partner and supplier negotiations. If the preceding three steps have been executed effectively, negotiations shift from adversarial bargaining to risk mitigation and ensuring the supplier receives fair compensation for the value delivered.

A “fair” cost baseline can be derived using robust bottom-up costing methods combined with an industry-average SG&A and profit markup. Forward-thinking executives use this stage to secure contractual covenants and performance clauses that protect the organization against cost or quality drift during the execution phase.

Manage—Priority set #5

Finally, execution and ongoing management are critical. Awarding business is only the beginning; disciplined follow-through is imperative. Utilizing weekly operational syncs to identify and resolve issues, tracking performance via supplier scorecards, and addressing deficiencies through structured Quarterly Business Reviews (QBRs) are vital practices. A strategy is only as effective as its execution.

Manifestation of our choices  

Cost management is a direct reflection of strategic choices. Top-performing organizations integrate cost planning across every value stream, establish clear priority sets, and focus on disciplined operational execution. Ultimately, the orchestration of these priority sets translates directly into the key performance metrics that define supply chain excellence: Cost, Quality, and Delivery.

About the author

Venkatesh Srinivasan is a supply chain and operations executive with over 20 years of experience at growth-oriented technology companies. His diverse experience covers automotive, consumer hardware, climate tech, and semiconductors. He is the VP of Global Supply Chain at a speciality materials company in the SF Bay Area. Views are personal.

Today is Friday, Oct. 30.

1. Walmart commits $6 billion to expand distribution automation

Walmart extended its Symbotic agreement, committing about $6 billion over several years to add automation at 25 U.S. regional distribution centers and support inventory management, storage capacity and replenishment globally.

Reuters  |   Read the original article

2. Freight spending reaches a five-year high despite flat demand

U.S. shipper spending rose 32.1% year over year in the third quarter while shipment volumes increased only 0.5%, as fuel costs and tighter driver availability pushed transportation expenses higher.

Reuters  |  Read the original article

3. Kroger moves closer to producers through direct sourcing

Kroger plans to increase directly sourced goods from less than 1% to more than 10%, building procurement hubs and import capabilities intended to improve supply visibility, costs and access to innovation.

Supply Chain Dive  |   Read the original article

4. Amazon absorbs higher freight costs while shortening delivery distances

Amazon said third-quarter shipping expense increased 11% as fuel and linehaul costs rose, although package consolidation, regional inventory placement and faster fulfillment helped the company continue reducing cost per unit.

FreightWaves  |   Read the original article

5. DoorDash acquisition would redraw grocery delivery competition

DoorDash agreed to acquire Instacart for $17 billion, combining broader delivery coverage with grocery fulfillment capabilities while leaving questions about retailer partnerships, platform choice and integration across the two networks.

Transport Topics  |   Read the original article

6. Record Southern California port volumes tighten inland capacity

Los Angeles and Long Beach set September cargo records as imports shifted toward the West Coast, increasing pressure on drayage, rail dwell and longer-haul trucking capacity during an extended peak season.

Transport Topics  |   Read the original article

7. Carrier capacity controls support container rates as demand cools

Container prices increased on major Asia-Europe and trans-Pacific routes as congestion and canceled sailings constrained available capacity, while returning Red Sea services and a possible West Coast labor disruption reshaped schedules.

FreightWaves  |    Read the original article

8. FedEx Freight faces technology costs in first standalone report

FedEx Freight reported higher revenue but lower adjusted earnings after its spinoff, as pricing gains were offset by transition expenses and monthly payments to its former parent for technology services.

Transport Topics  |   Read the original article

9. Intermodal growth lifts weekly U.S. rail traffic

U.S. rail traffic increased 7.8% for the week ended Oct. 24, with intermodal volume up 11.3% and eight of 10 carload commodity groups posting year-over-year gains.

FreightWaves  |   Read the original article

10. C.H. Robinson outlines integration approach for RXO acquisition

C.H. Robinson said customer overlap in its planned $5.8 billion RXO acquisition is manageable and expects combined data, pricing and last-mile capabilities to create synergies without replacing organic growth initiatives.

Transport Topics  |    Read the original article

Executive takeaways

Many direct-to-consumer brands, especially smaller Shopify, Amazon and TikTok Shop sellers, built their entire fulfillment model around shipping individual orders straight from Chinese factories or fulfillment centers to U.S. doorsteps.

William Yu

However, a major policy shift in the United States has upended global trade for these companies, forcing them to rethink their supply chains and even their overall business models because of the potential impact on their bottom lines. The conventional advice for importers is to find a U.S. fulfillment option to ship and store products domestically.

But there is another way. Brands don’t need to choose between China and U.S. fulfillment. These models are not mutually exclusive. They can be integrated into a fulfillment strategy that evolves with the brand. Before providing operational insights into this hybrid approach, let’s dig into how we got here.

De minimis changed the e-commerce supply chain

For decades, the U.S. has allowed certain low-value imports to enter the country duty-free and with simplified customs procedures. This “de minimis” treatment, mirrored in some form by many other economies, was intended to streamline and facilitate global trade.

In 2016, the U.S. raised its de minimis threshold from $200 to $800. The higher exemption helped enable retailers to redesign their supply chains around direct-to-consumer cross-border fulfillment. It also contributed to the growth of low-cost Chinese online retailers such as Temu and Shein.

The use of the de minimis exemption has exploded, growing from 224 million shipments in 2016 to more than 1 billion in 2023. A 2023 Congressional report found that Temu and Shein alone were likely responsible for nearly half of all de minimis shipments to the U.S. from China.

The Trump administration said the exemption was overused and abused, and suspended duty-free de minimis treatment for products from China and Hong Kong in May 2025, then extended the suspension to shipments from all countries effective Aug. 29, 2025. The suspension continued into 2026. The European Union followed with a similar change in July 2026, abolishing its €150 customs-duty exemption and introducing a temporary €3 customs duty per item on qualifying low-value imports.

The new math of cross-border fulfillment

The abrupt end of de minimis created chaos in cross-border e-commerce. Many industry observers initially predicted that the end of the exemption would signal the death knell for fulfillment from China. In reality, it simply transformed the math. The loss of duty-free status significantly reduced the tariff advantage of direct-to-consumer cross-border fulfillment, but it did not eliminate the structural advantages of manufacturing and staging in China and the deeply entrenched supply chains that fuel global trade.

The policy shift has forced DTC brands to tailor their supply chains to their growth stage:

• Early stage: Stay close to the factory. For brands just starting, the priority is minimizing inventory risk and preserving cash. By fulfilling orders directly from China, near the point of manufacture, businesses can avoid the heavy capital outlay required to pre-ship large quantities of bulk inventory to U.S. warehouses before demand is proven.
• Growth stage: Get closer to the customer. As a brand gains traction, the focus shifts to speed and consistency. At this stage, high-growth sellers should migrate to a distributed warehousing model in the U.S. Staging inventory domestically allows brands to offer the fast delivery speeds American consumers expect while generally calculating customs duties based on the value of the goods at import rather than the product’s eventual U.S. retail selling price. Under the standard transaction-value method, CBP generally bases the customs value on the price actually paid or payable for the imported goods, subject to applicable adjustments.
• Mature stage: Go global. Truly mature brands require global expansion capabilities. This stage calls for a network of fulfillment centers across key international markets to provide a seamless, localized experience wherever customers live.

Think about inventory velocity, not just geography

Brands can also segment their inventory based on turnover rates. Proven, high-velocity core products, the best-sellers that drive consistent sales, are routed in bulk to U.S. fulfillment centers. For these high-volume items, bulk ocean freight spreads out customs clearance fees, port fees, and tariffs across thousands of units. This brings the per-unit shipping cost down while enabling the fast domestic delivery speeds that U.S. consumers increasingly expect.

Staging in China near the production line still makes sense for long-tail items, new product launches and seasonal variations. For unproven inventory, storing products in U.S. warehouses can add significant storage, handling and labor costs before demand is established. If a new product launch underperforms, the brand can limit its exposure to the additional overseas shipping, domestic handling and long-term storage costs that would come with pre-positioning large quantities of inventory in the U.S. If the product is a breakout success, the brand can seamlessly transition the next manufacturing batch into a bulk ocean shipment to replenish U.S. fulfillment centers.

About the author

William Yu is the founder and CEO of NextSmartShip, serving over 2,000 DTC brands globally. Coming from a marketing background, he founded NSS after identifying a critical gap between China’s manufacturing ecosystem and the global fulfillment infrastructure for modern ecommerce brands. He brings a customer-first perspective to logistics—treating fulfillment as a growth engine, not just a cost center. Today, NextSmartShip operates 20+ fulfillment centers across 13 countries, helping brands optimize inventory, reduce risk and scale internationally.

.

Today is Thursday, Oct. 29.

1. U.S. delays robotics tariffs while opening trade negotiations

The administration found imported robotics and industrial machinery pose national-security risks but withheld immediate tariffs, giving U.S. manufacturers temporary relief while agencies negotiate with trading partners over the next 180 days.

Reuters  |  Read the original article

2. U.S. warning raises new risk for Strait of Hormuz traffic

U.S. authorities advised vessels without an operational need to avoid the Strait of Hormuz, adding uncertainty for energy and container shipping through a chokepoint already affected by attacks and elevated insurance costs.

Transport Topics  |  Read the original article

3. Rail bottleneck traps nearly 35,700 containers at Southern California ports

Nearly 35,700 import containers awaited on-dock rail service at Los Angeles and Long Beach, with equipment shortages and record volumes extending inland delays as a possible security-worker strike creates additional risk.

FreightWaves  |  Read the original article

4. Nvidia supplier agreements target AI infrastructure constraints

Nvidia reached capacity agreements with six suppliers covering substrates, components, packaging and testing, seeking to expand selected production as much as fifteenfold while shortening lead times for increasingly constrained AI systems.

Supply Chain Dive  |   Read the original article

5. AI-related restructuring cuts supply chain entry-level roles

Gartner reports that 47% of surveyed supply chains reduced staffing by an average 10.5% over two years, with two-thirds of respondents saying reductions were concentrated in entry-level positions.

Gartner  |   Read the original article

6. Paccar supplier constraints limit truck deliveries

Paccar delivered 40,100 trucks in the third quarter, below its 42,000-unit expectation, as suppliers struggled to match stronger demand; management expects performance to improve during the fourth quarter.

Transport Topics  |  Read the original article

7. FedEx raises outlook ahead of peak parcel season

FedEx increased its annual earnings and sales guidance after quarterly results, giving shippers a stronger demand signal before the holidays while the carrier continues reshaping its network and customer mix.

Transport Topics  |  Read the original article

8. Old Dominion tonnage turns positive after prolonged decline

Old Dominion reported September tonnage growth for the first time in more than two years, although management said a broader freight recovery still depends on stronger demand and easing energy costs.

FreightWaves  |  Read the original article

9. China urges Europe to manage supply chain dependencies through dialogue

China called on Germany and the European Union to address trade and supply chain tensions through consultation after Berlin advocated broader Asian market access and reduced dependence on concentrated suppliers.

Reuters  |   Read the original article

10. Contracts can strengthen critical-mineral continuity

A proposed contracting framework would help manufacturers prioritize exposed critical-mineral suppliers and establish clearer obligations for traceability, shortage allocation, continuity planning and reporting across sub-tier supply networks.

Supply Chain Management Review  |   Read the original article

Executive takeaways

When China restricted rare-earth exports in 2025, U.S. automakers idled plants within weeks. The disruption exposed a national vulnerability: America’s advanced-manufacturing base—and the economic and industrial security that rests on it—depends on a handful of foreign sources for critical minerals. A single country controls roughly 70% of mining, 85% to 90% of refining, and about 90% of rare-earth magnet production.

Washington has responded with tariffs, stockpiling, and billions in processing incentives—but these rebuild the physical supply chain over years. Meanwhile, a faster and largely overlooked lever sits inside every U.S. manufacturer’s existing supplier relationships: the contractual architecture that governs how supplier obligations, material origin, allocation priority, and continuity are defined, verified, and enforced.

Today’s risk is increasingly one of allocation, not absence. When material is scarce, who gets served first is decided by contract terms—not geology. Yet most supply agreements are silent on origin traceability, shortage-allocation priority, sub-tier visibility, and continuity. That silence is the blind spot. Here is a practical, non-proprietary way to close it.

Start by narrowing the scope

A typical manufacturer has 1,000+ direct suppliers, and critical minerals often sit five to ten tiers deep. Mapping every mineral across every supplier is impossible. Three steps produce a manageable priority list.

• Map products to minerals. Identify the products that matter most and the minerals they depend on—magnets on neodymium and dysprosium; batteries on lithium, nickel, and cobalt; power electronics on gallium and silicon metal.
• Rank minerals by exposure. Combine supply concentration, geopolitical risk, and limited domestic processing to isolate the minerals that most warrant protection.
• Plot suppliers on a 2×2 matrix. Position each supplier by revenue/business impact (vertical) and critical-mineral exposure (horizontal), as shown in Figure 1.

Figure 1 — Supplier positioning is illustrative; it varies by a company’s products and mineral exposure.

The top-right quadrant—high impact and high exposure—is where the full contractual treatment goes first. Suppliers with strong revenue weight but lower exposure get key resilience clauses; high-exposure, lower-spend suppliers get targeted clauses and monitoring; the rest proceed on standard terms. This keeps effort proportionate.

Turn each risk into a clause

Once the scope is set, the method is a direct mapping from each supply risk to the contractual control that addresses it. A short, general library of model clauses covers the recurring exposures:

• Origin & traceability. A certificate of origin plus batch-level chain-of-custody, so material can be traced from mine to delivery.
• Allocation & priority. A guaranteed minimum, and a commitment not to rank other buyers ahead of you during a shortage.
• Business continuity. Buffer or bonded contingency stock and a documented continuity plan.
• Sub-tier reporting & early warning. A periodic report of origin, single-source dependencies, and any threatened disruption.
• Audit rights and a refined force-majeure clause. Verify rather than trust, and prevent a supplier from claiming relief for a shortage its own avoidable concentration caused.

A resiliency scorecard

The clause library defines what a resilient agreement contains; a scorecard assesses how well a supplier can meet those terms. Score each dimension 1–5 and weight to a composite; a single critical gap—say, refusal to disclose origin—can outweigh an otherwise strong score.

Figure 2 — Resiliency scorecard weightings are illustrative; they may vary by a company’s products and mineral exposure.

Crucially, this score should carry real weight in the overall supplier scorecard—where critical-mineral resilience has traditionally gone unrewarded alongside cost, quality, and delivery.

The hard case: minerals you can’t see

The honest challenge is depth. A critical input like silicon metal may sit six or seven tiers below your only contract—with a Tier-1 assembler you can influence and a smelter you cannot even see (Figure 3). You can’t score or contract with the smelter directly.

Figure 3 — An illustrative multi-tier path showing how a critical mineral can sit far below a buyer’s only direct contract.

The answer is to change what you assess. Instead of scoring the deep-tier supplier, score your Tier-1 on its cascade capability—its ability to map, flow obligations down to, and report on the tiers beneath it. Two clauses do the work: sub-tier reporting cascades origin and early-warning duties down the chain through back-to-back provisions, and audit rights make that reporting verifiable. Neither depends on a direct contract with the smelter; both travel through the chain you do control.

A candid caveat matters here: contractual flow-down weakens with depth. The realistic goal is not perfect enforcement at Tier 7, but maximized visibility, early warning, and documented breakpoints—turning an invisible risk into a monitored one.

Why this scales

None of this requires new technology or capital. It uses the contracts a manufacturer already signs. This article distills a fuller, non-proprietary playbook—a risk-to-clause library, a supplier resiliency scorecard, and an implementation roadmap — intended for open adoption across the U.S. automotive and advanced-manufacturing sectors. Broad adoption of a shared contractual approach would strengthen supplier-obligation visibility and continuity across the national industrial base: a resilience gain that complements the physical build-out, and one that can begin at the very next contract renewal.

I welcome collaboration with procurement peers, trade associations, and standards bodies to refine and adopt this approach industry-wide.

About the author

Ashish Patil is developing and disseminating a non-proprietary contractual framework for critical-mineral supply-chain resilience for the U.S. automotive and advanced-manufacturing industries. He has two decades of experience in procurement, strategic sourcing, and supplier-risk governance across global manufacturing, and holds an MBA in Supply Chain Management from Michigan State University.

Today is Wednesday, Oct. 28.

1. Boeing production costs expose continuing supply constraints

Boeing reported a $116 million quarterly loss as higher commercial-airplane costs, production-rate adjustments and supplier pressures slowed its recovery despite stronger revenue and continued positive cash generation.

The Wall Street Journal  |  Read the original article

2. AI investment keeps U.S. equipment spending elevated

Core capital-goods orders rose 0.5% in September while shipments increased 1.2%, signaling continued business-equipment investment even as borrowing costs and questions about AI infrastructure returns create uncertainty.

Reuters  |   Read the original article

3. Storm damage cuts rail access to major U.K. ports

Storm damage and a derailment suspended or reduced rail freight serving London Gateway and Tilbury, pushing shippers toward road alternatives while repair timing, trucking capacity and container backlogs remain uncertain.

The Loadstar  |   Read the original article

4. UPS targets data-center logistics as a growth market

UPS estimates data-center construction could create a $15 billion annual revenue opportunity, supporting its shift toward specialized, higher-margin logistics as the carrier reduces lower-return Amazon package volume.

Reuters  |   Read the original article

5. Taiwan expects technology demand to sustain exports

Taiwan’s central bank expects AI and other emerging-technology demand to support exports and private investment through 2027, although growth should moderate and geopolitical and weather risks remain significant.

Reuters  |  Read the original article

6. Kroger connects store-level inventory signals with suppliers

Kroger is sharing computer-vision insights with suppliers and piloting RFID across roughly 1,500 store bakeries, aiming to improve availability, accelerate inventory counts and reduce waste in fresh categories.

Supply Chain Dive  |    Read the original article

7. AI construction tightens the market for outdoor storage

Data-center construction is increasing demand for industrial outdoor storage yards near ports and development hubs, while permitting limits and fragmented ownership constrain supply and push rents higher in major markets.

The Wall Street Journal  |  Read the original article

8. Tighter carrier vetting reshapes truckload capacity

Knight-Swift has reduced its approved brokerage carrier pool by about 70% over two years, citing cargo theft, safety and liability concerns as stricter vetting intersects with shrinking trucking capacity.

FreightWaves  |   Read the original article

9. Typhoon congestion temporarily supports container rates

Typhoon Champi and vessel bunching are delaying Asian port schedules and absorbing capacity, but excess supply, price competition and returning Suez services continue to weigh on longer-term container rates.

The Loadstar  |   Read the original article

10. A century-old mill offers lessons in supply resilience

The North Dakota State Mill illustrates how sustained capacity reinvestment, regional supplier relationships, diversified channels and employee incentives can support resilience while requiring careful management of concentration and demand risk.

Supply Chain Management Review  |  Read the original article

Executive takeaways

The North Dakota Mill and Elevator, commonly known as the North Dakota State Mill, has been operating since 1922. Originally built by the state of North Dakota to protect its farmers from what were considered predatory practices by Minneapolis-based railroads and milling facilities, the Mill has evolved considerably over the past 100 years. According to the Bank of North Dakota, its mission is to promote and provide support to North Dakota’s agriculture, commerce, and industry. It plays a vital role in the state and can offer insights useful to private enterprises and cooperatives nationwide.

The Mill has consistently focused on the long term, choosing to develop its facilities and increase capacity rather than chase short-term returns for its stakeholders. This has made it the largest single-site flour mill in the United States, according to the Grand Forks Herald, with 10 milling units that clean, process, and mill over 130,000 bushels of wheat daily. According to the Mill, it adds value to 40 million bushels of wheat annually.

Cooperatives and family-owned businesses can benefit from a similar approach, choosing to prioritize long-term results, especially in terms of increasing capacity, over short-term gains for their stakeholders.

The Mill focuses on its regional location and competes on the quality of local wheat. It is bolstered by its strong relationship with local farmers and grain elevators and benefits from North Dakota being among the largest wheat-producing states in the U.S. By collaborating and maintaining relationships with regional partners, while reducing supply chain distances, the Mill has quietly become a resilient institution that has weathered crises such as the COVID-19 pandemic, with a reported increase in home baking leading to increased demand for its Dakota Maid flour, as mentioned by the North Dakota Monitor.

Private enterprises would do well to emulate these practices, as supply chain resilience continues to grow in importance as firms continue to be exposed to a variety of exogenous shocks that can significantly impair operations.

The Mill has a diverse mix of products that it delivers to its customers via multiple channels. According to the Mill, 80% of its products are shipped via bulk rail cars and trucks, whereas 20% are shipped via packaged 5-, 10-, 25-, and 50-pound bags. Additionally, the Mill sells bread machine mixes and pancake mixes and holds organic certification for its wheat products.

An intentional approach to industry trends by focusing on organic certification and diversifying its sales channels has enabled the Mill to remain competitive. While some level of dependence on customers who purchase in bulk is the norm in the commodities sector, private enterprises would benefit from diversifying sources of revenue, increasing their resilience and reducing dependence on any individual client or partner.

From a human-resource perspective, the Mill is a robust example of an institution that has a substantial profit-sharing program for its employees. The Grand Forks Herald reported profit payouts averaging $27,000 for 143 of its 156 employees in fiscal year 2024. The Mill has shown its workers that they are an integral part of its operations and has made them de facto partners in the process. The 13 employees who did not participate in profit sharing that year had put in fewer than 1,000 hours of work at the Mill.

The Herald reported that the Mill’s total profits increased further in fiscal year 2025, to $26.7 million, with over $4.6 million paid to the Mill’s employees. Cooperatives and family enterprises can use similar strategies to show employees they are an integral part of the business, which may boost retention and morale and may help reduce training costs.

Finally, the Mill focuses on its state brand to highlight its history, its local roots, and connections to the state. As consumers show increased interest in domestic brands, this is an example of intelligent branding by the Mill. Private enterprises can utilize similar techniques, as has been done by cooperatives in other agricultural sectors in the U.S, such as Tillamook or Cabot in the dairy sector, by focusing on their homegrown brand and appealing to their consumer base.

According to the Mill, it receives no financial assistance from the state of North Dakota to subsidize its operations and has contributed over 50% of its profits to the North Dakota State General Fund for over 50 years. Despite this, the Mill continues to invest heavily in its capacity and its employees.

While detractors may argue that the Mill benefits from state ownership and a lack of shareholder pressure, private enterprise would not have to deal with some of the constraints the Mill faces, such as political pressure or expectations to transfer a substantial share of profits, which could increase funding available for reinvestment in long-term operations. Table I highlights the lessons supply chain leaders can learn from the Mill.

Table I: Lessons from the North Dakota Mill and Elevator

The Mill benefits from operational discipline it has maintained for over a century, including investing in capacity and employees, staying close to its raw materials, and building a state brand, while balancing bulk and packaged products.

Practitioners would benefit from adopting a written reinvestment policy, building long-term supplier relationships, developing multiple sales channels, and tying employee rewards to profit. While the Mill faces no takeover pressure, it does face political pressure from the state—but its operating practices can be replicated by cooperatives, family firms, or private businesses willing to adopt a long-term view.

About the authors

Omar Farooq is an Assistant Professor of Management in the School of Business and Economics at the University of Wisconsin, River Falls. His research investigates facility location, focusing on reshoring and nearshoring, operational efficiency, and supply chain resilience.

Irita Mishra is an Assistant Professor of Management in the School of Business and Economics at the University of Wisconsin, River Falls, where she teaches operations management, supply chain, and artificial intelligence for business. Her research examines procurement negotiation with algorithmic buyers, coordination between firms, and sustainability and risk in semiconductor supply chains.

Today is Tuesday, Oct. 27.

1. U.S. and Canada pursue relief from steel and auto tariffs

U.S. and Canadian officials are discussing possible relief from Section 232 steel and auto tariffs, but the privately conducted talks remain unresolved and could be affected by political timing.

Reuters  |    Read the original article

2. South Korean investment anchors U.S. shipbuilding ambitions

South Korean shipbuilders are committing capital and expertise to U.S. yards, including Philadelphia expansion and a proposed advanced facility, as Washington seeks to rebuild domestic commercial and naval capacity.

The Wall Street Journal  |   Read the original article

3. Mexican truckers strike over security and operating costs

Mexican trucking groups launched an indefinite national protest over cargo theft, extortion and rising costs, creating a potential disruption risk for domestic freight corridors and cross-border trade with the United States.

FreightWaves  |    Read the original article

4. USTR considers extending the pause on China vessel fees

The U.S. Trade Representative proposed extending its suspension of China-related maritime fees through Jan. 9, 2027, giving shippers additional near-term certainty while the agency collects public comments.

Logistics Management  |  Read the original article

5. UPS schedules a 5.9% average rate increase for 2027

UPS will raise average package rates by 5.9% on Dec. 21, continuing a multiyear pattern of general increases as shippers prepare for changes in base charges, surcharges and discounts.

Logistics Management  |   Read the original article

6. AI hardware becomes a higher-value cargo-theft target

Overhaul reports that thefts involving data-center equipment, chips and computing components have increased as criminals use fraudulent carrier identities and other deception to pursue valuable artificial-intelligence hardware shipments.

Supply Chain 247  |  Read the original article

7. USPS will add a weekly fuel surcharge in January

The Postal Service will introduce a weekly fuel surcharge for several domestic package services in January 2027, tying the percentage to federal diesel and jet-fuel price data and publishing updates electronically.

Supply Chain Dive  |   Read the original article

8. AI adoption outpaces operational redesign

A survey of more than 200 supply chain-intensive organizations found 69% had moved beyond isolated AI use, yet only 19% had substantially redesigned workflows to support human and AI collaboration.

DC Velocity  |   Read the original article

9. Connected reusable packaging expands shipment visibility

RFID, GPS, Bluetooth and sensor technologies are turning reusable totes, pallets and containers into trackable assets, helping operators monitor location, condition, utilization and return cycles across distribution networks.

DC Velocity  |   Read the original article

10. Freight recovery faces capacity and cost headwinds

U.S. truck tonnage grew modestly through September, but regulatory changes, fuel expenses and uneven demand continue to complicate the freight recovery and may influence capacity and contract negotiations.

Transport Topics  |   Read the original article

Executive takeaways

Inventory handling is becoming more complicated with the rise in the number of products, customers, and logistics issues. A company can have thousands of SKUs, with varying degrees of demand to manage. There are certain items that have good sales volume, while other items may sell occasionally; and there are products with unpredictable demand, and yet others that have not been sold for many months and may be sold quickly without prior warning. Since it is almost impossible to maintain a unique inventory policy for each SKU, a company may divide its products into different segments. Segmentation groups similar products together, thus allowing the management of the various groups in an easier way. One such valuable approach is the Runners, Repeaters, Strangers, Aliens (RRSA) classification.

To put it simply:

• Runners are goods with regular demand.
• Repeaters are goods with regular but not as frequent demand.
• Strangers are goods with irregular demand.
• Aliens are goods that are not sold at all or have been sold very little recently.

This classification helps make it easier to organize a complex inventory structure. However, it is observed that demand does not always remain constant. A Runner product can eventually turn into a Repeater, while a Stranger may become a regular supplier due to the new client or market evolution. Even a potential Alien may gain significance in case of an accident, product launch, or any disruption occurring in the supply chain. Therefore, the important question is no longer simply: “What category does this SKU belong to?”

It becomes “What category is this SKU likely to move into, and what should the company do about it now?” This shift is the foundation of the Adaptive Inventory Segmentation framework.

Why static inventory classification is not enough

Though traditional inventory management is useful for determining how a particular item may behave over time, predicting its future outcome can be less likely. For instance, consider a spare part for machinery that might have been sold sporadically over the years, merely a few times a year and thereby earning the label Stranger. And yet, one of the company’s major clients might suddenly start using that machine more frequently, which would cause the demand for that spare part to grow month by month. Despite this change, a traditional inventory classifying technique would not reclassify that spare part from the category of Stranger until enough data is gathered to do so. Meanwhile, the company could already be experiencing an understocking situation. The same applies to the opposite situation, when an item is classified as a Runner because there has been many sales of this item over the years, thus requiring ample inventory of it. However, in the event of changing preferences on the part of customers, the emergence of substitute products in the industry, or the expiration of the product’s life, the demand can gradually decline. This creates two possible issues:

• Understocking: The firm does not respond swiftly enough to a growing demand for one product.
• Overstocking: The firm continues to stock items with decreasing demand.

The issue arises not only from the absence of any classification system. The main issue that must be addressed is the limitation of a static system to identify any transitions between demand regimes. This leads to the main idea of the proposed framework.

Inventory categories should be treated as dynamic states rather than permanent labels.

The core idea: Demand-Regime Migration

The framework presents the idea of Demand-Regime Migration. Though the terminology sounds sophisticated, the underlying concept is not difficult. Demand-Regime Migration refers to the shift of an SKU from one type of demand to another. A product’s shift from being a Runner to a Repeater implies that it is experiencing progressively more intermittent demand. If demand shifts from being Stranger to Repeater, it indicates that demand is becoming more stable. If demand shifts from being Alien to Stranger, it means that a product previously classified as inactive is starting to gain traction amongst the customers. Hence, the new framework considers the following factors:

1. What is the status of SKU?
2. Where is it expected to move next?
3. What actions should the business take in relation to that expected change?

Demand characteristics → Current RRSA state → Transition probability → Inventory policy → Cost and service outcome

Before making assumptions of where an SKU is headed, the company must find out where it is now. The framework does not depend solely on product order frequency. Various characteristics of demand are considered, such as demand frequency, Coefficient of Variation (CV), and Average Demand Interval (ADI).

The important difference: Classification vs prediction

This represents the key difference between the proposed framework and traditional RRSA. In a conventional classification scheme, we would say, “This product is still a Repeater.” A more sophisticated adaptive framework asks, “This product is still a Repeater, but what is the likelihood of the product remaining a Repeater in three months?” That will give us a transition probability. Suppose the model estimates that the Repeater has the following probabilities:

• 68% chance of remaining as a Repeater
•  24% chance to become a Runner
• 8% chance to turn into a Stranger

The manager does not receive a single label. Instead, a picture of the SKU’s possible future emerges. This change in concept is very significant. The framework does not aim to forecast the future with precision; instead, it estimates the probabilities of its future states.

How machine learning fits into the framework

Machine learning is not intended to replace inventory management decisions; rather, it helps identify patterns that may signal a change in demand behavior. The process begins with historical SKU-level data, including demand frequency, average demand, recent demand changes, demand trend, seasonality, historical sales, lead-time variability, annual consumption value, and criticality. The model then examines how these characteristics were associated with changes in RRSA status in historical observations. For example, SKUs showing increasing demand frequency, a positive demand trend, and shorter intervals between purchases may have been more likely to move from Stranger to Repeater. Conversely, SKUs showing declining demand, longer intervals between purchases, and persistently low demand may have been more likely to move from Repeater to Stranger. By learning these relationships from historical examples, the model can estimate the probability of future transitions for a currently classified SKU.

An illustrative example:

Assume that a company has historical records for 5,000 SKUs. For each SKU, the company has information on demand behavior over a six-month period and observes the RRSA status of the SKU over the subsequent three months. These historical observations provide examples of how different demand characteristics are associated with changes between RRSA categories.

The machine-learning model uses these examples to learn relationships between the characteristics of an SKU and its subsequent RRSA transition. For instance, the model may identify that Repeaters with increasing demand frequency and a positive demand trend were more likely to become Runners in the following three months.

Now consider a new SKU that is currently classified as a Repeater. Its latest demand characteristics are provided to the trained model. Based on the patterns learned from the historical data, the model may estimate a 72% probability that the SKU will transition from Repeater to Runner within the next three months.

The 72% probability does not mean that the SKU will become a Runner. It indicates that, based on the historical relationships learned by the model, the estimated likelihood of this transition is 72% within the specified prediction period. This provides the manager with information about transition risk rather than only the current RRSA classification. Therefore, the model helps the company prepare for a potential change before the SKU is formally reclassified.

Why machine learning is useful here

Classic methods of demand classification typically depend on rules that have already been determined. For instance, “The SKU is considered as Runner if demand frequency is over a specific threshold.” These rules can be useful but cannot handle situations where various factors interact. Assuming that a product is becoming more frequently used, the lead time is likely to be uncertain, it is crucial for operation, and its annual consumption value is high, an Eight-box rule will not allow the recognition of the complexity of the situation. Machine learning algorithms such as Random Forest and Gradient Boosting can analyze several levels at once and find tendencies that cannot be revealed using a single rule. This article is aimed at assessing the effectiveness of these methods in comparison with traditional logistic or multinomial regressions. It is vital that the model is verified out-of-sample through a time-based validation that rules out the unintentional use of the future data for predicting the past.

Train using older data → Predict a later period → Compare prediction with what happened. This gives a more realistic test of whether the model can support future inventory decisions.

From prediction to action

Prediction, in and of itself, does not lead to lower inventory costs. Its value lies in altering the inventory decision. This is where the field of Operations Research comes into play. Assume that the model indicates a significant probability for a specific product to be a Runner, the decision may involve increasing safety stock, increasing the frequency of replenishment, increasing availability, or ensuring supplier capacity. Conversely, if the model predicts the product is likely to be a Stranger or an Alien, possible actions may be to reduce cycle stock, decrease replenishment frequency, consider make-to-order system, minimize purchasing risks, or exit the market altogether. However, there is one parameter that can change everything: Criticality. A low-demand component for a safety-critical machine cannot be put in the category of an ordinary low-demand product. For example, if some spare part has demand only twice a year, one could argue that it is classified as an Alien in terms of demand. However, if this part is necessary for a multi-million-rupees production line, keeping some stock could be economically plausible.

Low demand does not automatically mean low importance. Therefore, the framework explicitly incorporates criticality into the policy response.

A practical example: Why two repeaters may need different policies

Consider the implications of two SKUs, namely SKU A and SKU B. They are both presently categorized as Repeaters. This means that the conventional system would use the same replenishment scheme for these two SKUs. However, predictive analysis provides another perspective.

SKU A’s details are as follows:

• Current Status: Repeater
• Probability of becoming Runner: 76%
• Lead Time: Stable
• Priority of Customer: Very Important

This data indicates that SKU A is entering a period of regularity, and perhaps a higher demanding period. If it is not treated as more than just another Repeater unit, the company may face unexpected stock shortages. Instead, an adaptive approach should be taken in order to adequately protect the unit and ensure effective replenishment.

SKU B’s data is as follows:

• Current Status: Repeater
• Probability of becoming Stranger: 79%
• Demand: Declining
• Holding Cost: High

SKU B is moving in the opposite direction. The declining demand and high cost of holding inventory imply that the same policies concerning SKU A inventory cannot be applied to SKU B. The key insight is that both products are currently Repeaters, but they are moving in different directions.

This is the practical value of adaptive segmentation. The company does not wait until SKU A officially becomes a Runner or SKU B officially becomes a Stranger. It starts preparing for the change before the classification changes.

Measuring whether the framework actually works

One of the key ideas presented in the current research is that a better predictive model does not at the same time mean a better inventory system. For instance, if we have a machine learning model that provides 10% better accuracy in classification, this is impressive, but without changing the inventory policy or making the necessary changes to the inventory system, it will hardly have an operational value. This is precisely why the framework assesses the business results, which may include total inventory cost, average inventory, stockout ratio, service level, inventory turnover, and obsolete inventory. This allows researchers to evaluate the effectiveness of their results not only from the perspective of whether they predicted correctly.

Testing the framework under different conditions

Actual supply chains function with many kinds of demand at their disposal. Demand can be constant, seasonal, sporadic, unstable, increasing, decreasing, or based on unexpected events. Hence, the framework suggests the implementation of Monte Carlo simulation in creating different scenarios to analyze how the adaptive mechanism performs under different conditions. For instance, one simulation may show a case of stable demand, whereas the other one may imply unpredictable demand with long lead-times. Lead-time uncertainty can vary as well to determine at what point adaptive segmentation brings the most value.

What should it be compared against?

To demonstrate that the framework provides genuine additional value, it needs to be compared with existing approaches such as ABC Classification, ABC–XYZ Classification, Static RRSA, Machine-learning forecasting combined with traditional replenishment, and Adaptive inventory segmentation. The comparison should focus on operational outcomes rather than simply asking which algorithm has the highest prediction accuracy.

Lower total inventory cost + same or better service level + fewer stockouts + higher inventory turnover + lower obsolete inventory.

Avoiding the ‘false improvement’ problem

Inventory optimization always involves trade-offs. The example may illustrate the situation when a firm decides to reduce its inventory level by 20%. At first sight, it seems to be a great achievement. However, let’s assume that stockouts increase by 15% at the same moment. Thus, it may be argued that the reduction of inventory fails to show real performance improvement because customer service level has dropped significantly. Similarly, the reduction of inventory cost by 8% should also be supported by statistical data and evidence proving the practical significance of such a reduction. A negligible change of 0.5% may not really change anything when it comes to practical aspects, while a 5% to 10% change in thousands of SKUs will have substantial monetary meaning. For this reason, the current article argues that it is essential to pay attention to balanced performance instead of working just on a particular indicator.

What this means for supply chain managers

A key change in management is the shift from classification management to risk intervention. Instead of a dashboard showing something like this: “SKU 1045 – Runner,” it could provide information such as: SKU 1045 – Repeater, 72% likelihood to become Runner within three months, High criticality. Action to take: increase the readiness for restocking.

It could also be such as SKU 2078 – Repeater, 79% likelihood to turn into Stranger, High holding cost. Action to take: Reduce inventory exposure. This changes the manager’s question. Instead of asking, “What category does this SKU belong to?”, the manager asks, “Which SKUs are approaching a meaningful change, how confident are we about that change, and what is the most cost-effective response?” The proposed framework therefore moves inventory management from classification control to proactive intervention.

Where the framework may not work as well

This method may not yield similar results in every case, including:

• When there is a minor number of precedent incidents.
• Demand is primarily event-driven, with limited comparable historical observations, making it difficult for the model to learn reliable transition patterns.
• Management is unable to determine product importance properly.
• Manufacturing time is very limited.
• Stockholding practices are dictated by law or by contractual obligations.
• Demand is so capricious that forecasting is very difficult.

In such circumstances, the unknowns of forecasting must be included in the decision-making process. For instance, a forecast may indicate a 55% chance of becoming a Runner and a 45% chance of remaining a Repeater. That is unlike 95% chance of becoming a Runner.

Consequently, credible decision-making systems must provide not only forecasts but also rate their reliability as forecasts.

The road ahead

The proposed framework opens numerous avenues for future exploration.

• Issues in multi-echelon inventory management (where the same SKU is subjected to different demands at different locations—factory, distribution centre, customers, etc.).
• Digital twin, wherein anticipated demand changes can be verified in simulated supply chain disruptions.
• Reinforcement learning, where an AI agent can learn its replenishment decisions from simulated realities multiple times.
• Sustainability, which includes the impact of the carbon footprint, transport, disposal costs, etc. and contamination from excess inventory.
• Humans and AI together making decisions, where the system can show the reasoning behind its decision as well as assess its cost.

The major aim for future implementations is to be adaptive through the processes of demand identification, recognition of demand regime change, forecasting of transitional processes, evaluation of alternatives, decision optimization, result recording, etc.

From knowing what inventory is to understanding what it is becoming

Figure 1. Adaptive Inventory Segmentation Framework.

In the past, inventory management has paid attention only to past performance of the SKU. The classification methodologies like RRSA serve to achieve this objective. However, demand is not constant, and the past behavior of the product does not provide a reliable clue about its performance in the future. The Adaptive Inventory Segmentation framework proposes a different approach. The key breakthrough, thus, is not just introducing machine-learning technologies into inventory management. It is establishing the link between the prediction of data and actual actions of the company. An SKU classified as Repeater, but likely to become a Runner, should not necessarily be treated in the same way as another Repeater classified as likely to become a Stranger.

Inventory should not only be classified according to what it has been. It should be managed according to what the data indicates it is becoming. This is the central proposition of the framework and the foundation for moving from static inventory classification toward continuously adaptive inventory management.

About the authors

Samriddha Basu, is an MBA grad in Operations and Supply Chain Management at the Symbiosis Institute of Operations Management, Nashik, and a Computer Science Engineering graduate. His interests and experience span supply chain management, manufacturing operations, production planning, analytics, and digital transformation. He has worked with organizations including CEAT Limited, JBM Auto, and PhyFarm on technology-enabled operational improvement initiatives. He holds certifications in CAPM®️, SAP Integrated Business Planning (IBP), Kinaxis Maestro Author Level 1, and Oracle Fusion Cloud Applications SCM & ERP Process Essentials, and is a graduate of the McKinsey Forward Program.

His project experience includes production scheduling and Lean process optimization at CEAT Limited, where he addressed material and scheduling mismatches across 152 SKUs using tools including Value Stream Mapping, Kanban, SMED, and digital dashboards; and manufacturing analytics at JBM Auto, where he developed a real-time MIS for operational and OEE visibility. His earlier technology work involved analytics, forecasting, and IoT-enabled workflows across 347 connected devices. He is particularly interested in the convergence of artificial intelligence, supply chain analytics, and operational excellence, with a focus on translating emerging technologies into scalable and practical solutions for complex supply-chain and manufacturing environments.

Dr. Deepak Datta Nirmal is an Assistant Professor at the Symbiosis Institute of Operations Management (SIOM), Symbiosis International (Deemed University), India. He earned his Ph.D. in Operations and Supply Chain Management from the Indian Institute of Management (IIM) Jammu. His research interests span Supply Chain Management, Industry 4.0, sustainable supply chains, and the application of advanced analytical approaches to contemporary industrial challenges. His research has been published in leading international journals, including Technological Forecasting and Social Change, Annals of Operations Research, and Benchmarking: An International Journal.

Dr. Nirmal has a strong focus on industry-oriented learning and has mentored students across multiple live industry projects addressing real-world challenges in operations and supply chain management. His mentoring engagements have covered areas such as inventory optimization, supply chain transformation and network design, AI-enabled demand forecasting and planning, production process improvement, and waste minimization. He is particularly interested in developing industry-based case studies that capture operational challenges, transformation journeys, innovative practices, and organizational success stories, while using analytical and case-based methodologies to derive actionable insights. He recently developed the Supply Chain and Operations Knowledge Platform, a free knowledge-sharing platform created to facilitate the exchange of knowledge among students, practitioners, and academicians and to bridge the gap between academic research and industry practice.

Ramakrishna (Ram) Garine is a supply chain analytics and AI practitioner and Thought leader, focused on disruption planning, simulation, and decision intelligence. His work spans applied machine learning for forecasting and risk, resilience metrics, and practical methods to evaluate mitigation policies such as alternate routing, safety stock, and supplier diversification. Ram contributes to professional communities through speaking, peer review, and research dissemination, and he builds tools and frameworks that make advanced resilience testing accessible to both industry teams and learners.

Today is Monday, Oct. 26

1. Supplier delays deepen as U.S. demand stays strong

S&P Global’s October survey showed manufacturing strengthening while supplier delivery times deteriorated to a four-year low, as precautionary inventory building, shipping delays and capacity constraints increased backlogs and pricing pressure.

Reuters  |  Read the original article

2. Energy attacks push oil to a six-month high

Brent and U.S. crude posted their highest settlements in nearly six months as attacks on Saudi infrastructure, maritime-security incidents and an oil spill intensified concern about physical supply and transportation risk.

Reuters  |  Read the original article

3. Canada tariff revisions redraw sourcing exposure

The White House will remove plywood, selected machinery and certain packaging materials from 50% Canada tariffs on Oct. 30 while adding cocoa powder, carpets, air conditioners and other products.

Supply Chain Dive  |   Read the original article

4. Freight indexes point to broader transportation cost pressure

The TD Cowen/AFS Freight Index projects higher fourth-quarter parcel, less-than-truckload and truckload costs, with fuel surcharges and carrier pricing changes adding pressure as shippers enter peak season.

Logistics Management  |   Read the original article

5. Fuel prices become trucking’s leading industry concern

The American Transportation Research Institute’s annual survey ranked fuel prices as trucking’s top issue after a 17-place jump, while insurance costs, the economy and truck parking also remained prominent concerns.

Logistics Management  |   Read the original article

6. Massachusetts lawmakers seek a pause on Amazon drone delivery

Sixteen Massachusetts legislators asked the governor to freeze state approvals for commercial drone delivery, citing privacy, safety, labor and community concerns as Amazon advances deployment plans in the state.

Supply Chain 247  |   Read the original article

7. Medtech companies build resilience around chips and scenarios

Medical-device executives described using additional inventory, scenario planning and closer semiconductor-supplier relationships to prepare for disruptions while balancing product availability, regulatory requirements and the cost of redundancy.

Supply Chain Dive  |    Read the original article

8. Truckload contract increases hold above 10%

A new SONAR index indicates truckload carriers are securing contract-rate increases above 10%, signaling that tighter capacity and the freight-market recovery are strengthening carrier leverage in shipper negotiations.

FreightWaves  |    Read the original article

9. Container rates rise as carriers trim capacity

Drewry’s World Container Index increased 2% to $4,585 per 40-foot container, supported by stronger U.S. import demand and carrier capacity reductions on trans-Pacific and Asia-Europe services.

FreightWaves  |   Read the original article

10. Skills gaps keep supply chain AI pilots from scaling

ASCM research found AI adoption across 13 supply chain functions averaging 1.94 on a five-point scale, with process design, workforce readiness and change management limiting progress beyond pilots.

FreightWaves  |   Read the original article

Executive takeaways

Cisco’s demand planners are getting time back from an AI agent. The harder question is what the company does with that time and what it takes to extend the same approach across more complex supply chain workflows.

Speaking at the recent ASCM CHAINge conference in Long Beach, California, Joe McMorrow, who leads supply chain transformation at Cisco, said a production demand-planning agent is saving each of the company’s 42 demand planners at least six hours a week on forecast-versus-actual analysis.

The agent delivers analysis automatically and allows planners to ask follow-up questions through a conversational interface. Cisco has not used the savings to cut demand-planning headcount, though, McMorrow said. Instead, the company is redirecting that time toward better planning insights and more engagement with sales.

That result provides a concrete starting point for a transformation that remains a work in progress. Cisco is pursuing larger, connected workflows, but McMorrow described substantial challenges in capturing business knowledge, connecting fragmented data and maintaining the applications it builds.

“We’re figuring it out as we go,” he said.

McMorrow said Cisco’s approach has changed as the technology and the company’s understanding of it have evolved. His presentation offered a view of AI implementation that separates a working tool from the broader ambition of changing how supply chain decisions get made.

Move beyond personal productivity

Cisco is approaching AI at two levels. Personal agents can help employees find information, summarize meetings and draft material. Domain-specific agents address standardized work within functions such as demand planning, manufacturing and quality.

A useful personal assistant does not automatically change an operating process,” McMorrow noted. For the transformation team, the work starts with identifying activities, recurring questions and exceptions within a business domain, then determining where an agent can make that work more effective.

The demand-planning agent illustrates that approach. It targets time spent gathering and comparing information, giving planners a faster starting point for analysis. McMorrow described a longer-term vision in which personal agents interact with trusted domain agents to answer questions about forecasts, actual demand and the effects of scenarios on lead times.

That connected environment is a goal. It also introduces questions about access permissions, data security and workflow management that must be resolved as the system grows.

Cisco’s other priorities include assessing the supply chain impact of quality problems, evaluating whether large customer deals can be supported and handling the exceptions that accompany order changes. Each involves several functions and multiple sources of information.

Rather than attempting to automate everything at once, McMorrow said Cisco is selecting scenarios tied to major business drivers and using them to demonstrate what is possible while building the foundation for broader use.

Data cannot replace missing business knowledge

One of the biggest barriers is giving agents the business context needed to connect information and interpret it correctly.

A process library can describe a workflow. A data catalog can identify fields and sources. Neither necessarily captures the knowledge employees use to navigate an exception, determine which information applies or identify the team that needs to make a decision. McMorrow described that business knowledge as a major hurdle in connecting agents. Cisco’s experience also shows that some information gaps originate in the underlying process itself.

Quality containment offers an example. Once engineers isolate a problem to a component batch or date-code range, the company must determine where affected material and products sit across suppliers, factories, warehouses, service depots and customers. McMorrow said assembling that picture can take months because teams maintain separate data environments.

An agent could help collect the information needed to assess the impact. But Cisco encountered a more basic limitation involving engineering changes and deviations: Available records did not always establish whether a factory had actually used a revised part. Teams still had to contact the factory to find out.

Keep the systems that supply the rules

Cisco is building an automation layer over its existing applications, McMorrow said. Its approach retains enterprise resource planning, advanced planning and other systems that contain established business rules and deterministic calculations.

The goal is to connect information across functional silos and streamline work that crosses organizational boundaries. Existing applications provide rules and guardrails that Cisco does not want to recreate.

McMorrow also cautioned against assuming large language models can supply every answer. Cisco combines them with other tools and analytical methods, depending on the task.

“And so LLMs, you can build good reasoning models with them, but when you need the real answer, you need deterministic math behind it,” he noted.

Human review remains central. For example, a factory debug agent can recommend a cause of failure, but a person must verify that recommendation. McMorrow distinguished agents that gather information from those that recommend a decision and, eventually, those that take action autonomously.

Asked about agents independently correcting problems and writing changes back to source systems, he was explicit about Cisco’s position at the time of the presentation. “We’re not there,” he said.

Turn reclaimed time into an operating outcome

McMorrow acknowledged the difficulty of translating time savings into profit-and-loss results. Inventory, for example, responds to many influences, making it difficult to isolate the contribution of a particular tool.

Cisco’s approach includes commitments from business leaders to adopt the tools and improve the work. The questions include whether employees are actually getting the reported time back and whether leaders can explain how the process has improved.

For demand planning, the intended use of that time is clear: Spend less effort assembling basic information and more effort developing a better plan. Whether that produces the expected operating benefits requires attention beyond the initial deployment.

That distinction gives executives a useful way to assess AI proposals. A project should identify the task it changes, the time or effort it releases, and the work employees will do with the capacity created.

Plan for ownership after deployment

Cisco is also developing an operating model for building and sustaining agents. Employees can use personal tools, more advanced users can prototype domain agents, and professional teams can turn promising applications into trusted production tools.

McMorrow said effective contributors combine business-process knowledge with technical skills. Understanding the work, the data and the questions users need answered is part of building an effective agent.

Maintenance is another part of the equation. As users request new capabilities, agents may require additional data, logic and ongoing support. McMorrow relayed a reminder from Cisco’s IT partners: “every one of these things is an application and it’s going to require maintenance.”

Domain leaders ultimately remain accountable for whether the tools are used and effective, he said. That puts responsibility for the business outcome alongside responsibility for developing and supporting the technology.

For organizations beginning the journey, McMorrow recommended focusing on a high-value task and developing internal capabilities over time. “Start small, work on something that’s task-based and maybe more deterministic to start,” he said.

Executive takeaways

Editor’s Note: The views and opinions expressed in this article are solely those of the authors and do not reflect the views of Dell Technologies or any of its affiliates.

In early 2024, thousands of Porsche, Bentley, and Audi vehicles were stopped at U.S. ports because a small electronic component was linked to a supplier on the Uyghur Forced Labor Prevention Act (UFLPA) Entity List. The problem did not originate with Volkswagen’s direct suppliers, according to reporting from Reuters. It emerged several tiers upstream. Yet the consequences landed on finished vehicles waiting for delivery.

Incidents like this highlight a growing challenge for supply chain leaders: identifying whether a flagged supplier actually affects the products they buy, especially when obtaining verification data is becoming more difficult. As regulators increasingly expect product-level traceability, many of the tools traditionally used to obtain upstream sourcing information are becoming harder to access or less reliable. Manufacturers use third-party mapping tools, public records and supplier disclosures to identify potential upstream relationships. These sources support initial risk assessment, but establishing whether a connection affects purchased products requires further validation against internal records and supplier evidence.

The result is a new compliance dilemma: How should companies investigate a potential connection when exposure is uncertain and verification difficult?

The Volkswagen case was not an isolated event, and electronics carry most of the value at stake. As Table 1 shows, electronics represented 16% of shipments stopped under UFLPA but 82% of their value ($3.4 billion). By the reporting date, 59% of those electronics shipments had been released. A stop is not always a final denial, but importers must be ready to document where their products’ materials originated from, while the related supply chain operations come to halt or face higher uncertainties, leading to even higher costs. Verification delays can interrupt operations, increase uncertainty and raise costs. A held component shipment can affect production and distribution across several product lines, regardless of its value.

TABLE 1

Electronics represented 16% of UFLPA-stopped shipments but 82% of their value.

Coverage: June 2022–August 2026; data snapshot dated September 4, 2026. Outcome percentages use the stopped shipment count within each column. Values are rounded; outcomes reflect status at the reporting date.

Sources: CBP UFLPA Enforcement Statistics, Gatewell Group’s reproduction of CBP data.

What managers need to know

1. A flagged supplier does not automatically mean your products are affected.
2. Product exposure and verification difficulty are two different assessments.
3. Supplier disclosures often identify potential concerns without proving product-level involvement.
4. Limited visibility should trigger escalation, not reduce concern.
5. Effective investigations focus on specific products, materials, and time periods.

Obtaining that product-level evidence has also become harder. The Responsible Minerals Initiative (RMI) maintains the Conflict Minerals Reporting Template (CMRT), which companies use to exchange sourcing information on tin, tantalum, tungsten, and gold. In August 2026, China’s Ministry of Commerce (MOFCOM) issued countermeasures under Order No. 2, barring organizations in China from working with six entities, including the Responsible Business Alliance, which administers these programs, and several supply chain mapping and material testing providers. RMI has since released unbranded templates, so options remain. However, buyers now need to determine which channels remain open. (RBA statement; RMI template update)

These developments leave buyers with a practical question: how should they investigate a potential connection when product exposure is uncertain and supporting information is difficult to obtain? Consider Company A, a hypothetical manufacturer buying gold-plated connectors for servers from Supplier B. Public information links B to Refiner C, a UFLPA-listed entity, without establishing whether C’s gold enters A’s connectors. Using this example, the article explains how buyers can assess product relevance, focus limited investigative resources, avoid duplicate supplier requests and decide which unresolved concerns need escalation.

First determine exposure, then determine evidence availability

When information links a supplier to an entity on a restricted-party list, the instinct is to treat it as a single problem. In reality, two separate assessments are required: does the flagged material enter the products we purchase, and how difficult will it be to verify that?

Product exposure comes first. The question is whether Refiner C’s gold reaches the components Supplier B sells to Company A, or feeds a different product line, customer, or geography. That depends on internal records: what materials are specified, which entities appear in sourcing disclosures, and whether the product, purchasing entity, and time period align with the flagged connection.

A Financial Times review found that more than 160 U.S. companies could not rule out sanctioned smelters in their supply chains. Such disclosures identify potential upstream sources without establishing direct purchases or product-level involvement (Financial Times reporting). Amazon’s filing explains why the question stays open: suppliers reported facility information for their entire supply chains without specifying which facilities contributed metal to the products in scope, leaving the company unable to validate which of them “in fact processed the gold, tin, tungsten, or tantalum used in the in-scope products.” This example concerns sanctions screening rather than UFLPA enforcement, but illustrates a similar information gap. A company-wide CMRT may identify possible refiners without establishing which supplied materials for the buyer’s products. A practitioner involved in supplier mapping described the practical limit as follows: “A supplier map can show a real company relationship without telling us whether that material enters the product we buy.”

Verification difficulty is a separate issue. A high-exposure connection may be easy to verify if the supplier maintains clear records. A low-exposure connection may be nearly impossible to verify because upstream visibility is limited, records may not exist, or legal constraints restrict the supplier's participation in verification programs.

The mistake is letting one drive the other. When information is hard to obtain, the concern tends to receive less attention, not because exposure is lower, but because the path to an answer is unclear. Difficulty obtaining evidence should not lower assessed exposure; credible product concerns with inaccessible records warrant earlier escalation.

Choosing the next step

For Company A, the next step depends on what the connection reveals about its product. A company-wide disclosure requires clarification about which products it covers; a connection to the connector line requires records showing whether the refiner supplied gold during the purchasing period. The matrix combines these questions with the difficulty of obtaining evidence to guide the investigation.

 

These hypothetical examples address unresolved connections for a defined product and period. “High” and “moderate” describe the strength of the available product connection, not measured probabilities. Cases with insufficient information need initial scoping; document supported exclusions, and confirm involvement through compliance review.

The company-wide disclosure scenario reflects a limitation described by TE Connectivity, a manufacturer of connectors and sensors. In its 2025 conflict-minerals report, the company explains that broad supplier responses can leave it unable to determine which reported mineral sources relate to the parts it purchases. Although this concerns a separate reporting requirement, it illustrates why a supplier disclosure may require further product-specific investigation. (TE Connectivity, 2025 report)

For UFLPA reviews, CBP (Customs and Border Protection) identifies purchasing, inventory, and production records among the evidence used to trace supply chains. Once the buyer has identified what remains unresolved, the next step is to ask the supplier for records that address that specific question. (CBP Operational Guidance for Importers, pp. 13–15)

Four rules for supplier investigations

Rule 1: Ask a product-specific question

“Does material from Entity C enter the products supplied to us during the period under review?”

Rule 2: Share the reason for the inquiry

Tell suppliers why the concern exists and what needs verification.

Rule 3: Evaluate the evidence, not the assertion

Review supporting records rather than relying on broad supplier declarations.

Rule 4: Document unresolved gaps

Identify what remains unknown and assign ownership for follow-up.

Once the buyer selects an investigation approach, the supplier request should address the specific gap. For the connector example, Company A could ask: “For the connector part numbers and purchasing period identified, does gold from Refiner C enter the products you supply to us? What records support your answer?” Sharing the source of the flagged connection helps Supplier B understand what needs investigation. As one practitioner involved in supplier investigations put it, “The more homework we do on the likely product and material, the easier it is for procurement to help us.”

The request should also allow the supplier to explain barriers to obtaining information. Microchip Technology, a semiconductor manufacturer, describes one such barrier in its 2024 conflict-minerals report: some upstream suppliers declined information requests because Microchip was not their direct customer. The company reported working more closely with its contract manufacturers to bridge that gap, while acknowledging that difficulties remained. For Company A, the practical lesson is to work through Supplier B’s existing upstream relationships and agree on follow-up, rather than assume an incomplete answer means unwillingness to cooperate. (Microchip Technology, 2024 report).

Suppose Supplier B says Refiner C serves a different product line and provides a current sourcing list for the connectors. Company A must check whether that list covers the purchasing period under review. If it reflects only a recent sourcing change, it does not resolve the earlier period. Company A should request records covering the earlier period rather than another general declaration. Without those records, the response leaves the earlier purchases unresolved.

This example illustrates how to evaluate a response: distinguish what the evidence supports from what still needs verification. Reuse records that cover the relevant entity, products and period, and broaden the inquiry when new information reveals additional sourcing paths or contradictions. The checklists below turn this approach into repeatable steps.

Two practical checklists

Product exposure checklist

• Verify the source behind the alert.
• Identify affected entities, materials, and dates.
• Compare them against purchasing records.
• Determine whether the connection reaches your products.
• Document supported exclusions and assumptions.

Supplier verification checklist

• Ask a specific product-level question.
• Request supporting records.
• Verify coverage of products and time periods.
• Resolve contradictions and information gaps.
• Record decisions, ownership, and next action

Key lessons learned

1. A supplier disclosure can identify a concern without resolving it. TE Connectivity’s experience shows why a list of possible refiners may still leave product exposure uncertain. Before using a response to close an inquiry, buyers should check that it covers the relevant products and purchasing period and includes evidence supporting the conclusion.
2. Difficulty obtaining information should change the investigation approach, not lower the assessed exposure. Microchip’s work through contract manufacturers illustrates an alternative route to upstream information, although gaps remain. Buyers should identify the barrier, involve the relevant supplier contacts, and agree on follow-up. Procurement and compliance should decide how to address credible product concerns that remain unresolved.
3. Information must reach someone responsible for acting on it. The Senate investigation found that automakers were notified about JWD components in January 2024. Volkswagen disclosed affected vehicles that month, while BMW continued importing products containing JWD components until at least April, according to the committee. The case reinforces the importance of linking supplier warnings to affected products, assigning responsibility, and ensuring action follows.

Final takeaway

Enforcement agencies increasingly expect companies to understand not only who their suppliers are, but whether specific materials from upstream entities enter specific products. At the same time, obtaining product-level sourcing evidence is becoming more challenging.

Organizations that separate exposure assessment from verification difficulty can focus resources where the risk is greatest, escalate unresolved concerns sooner, and avoid treating supplier disclosures as definitive answers.

The goal is not perfect visibility. The goal is to make defensible decisions based on the best available evidence while clearly identifying what remains unknown.

About the authors

Andres Cuellar (corresponding author) works in supply chain intelligence at Dell Technologies. He can be reached at Andres_Cuellar@Dell.com.

Dr. Kefeng Xu is Professor of Supply Chain Management in the Dept. of Information Systems & Supply Chain Management at the Bryan School of Business & Economics at the University of North Carolina Greensboro in Greensboro, NC. He can be reached at Kefeng.Xu@uncg.edu.

Dr. Jayanth Jayaram is the Earl Congdon Distinguished Professor of Supply Chain Management in the Congdon School of Supply Chain Management at the University of North Carolina at Wilmington, NC. He can be reached at jayaramj@uncw.edu.

Dr. Vashkar Ghosh is Associate Professor of Supply Chain Management in the Dept. of Information Systems & Supply Chain Management at the Bryan School of Business & Economics at the University of North Carolina Greensboro in Greensboro, NC. He can be reached at V_GHOSH@uncg.edu.

Sources and related reading

Enforcement data and guidance

• U.S. Customs and Border Protection: UFLPA enforcement statistics. Data supporting Table 1.
• U.S. Customs and Border Protection: Operational Guidance for Importers, pp. 13–15. Supply-chain tracing and supporting documentation.
• U.S. Senate Finance Committee: 2024 investigation into automakers’ supply chains. Evidence supporting the automotive example and lessons on responding to supplier warnings.
• Shepardson, David (2024). BMW imported 8,000 vehicles into the U.S. with parts from banned Chinese supplier, Senate report says. Reuters; republished by Investing.com.

Company disclosures

These reports concern conflict-minerals reporting, a separate requirement from UFLPA. They illustrate the practical limits of supplier information.

• Amazon: 2025 Conflict Minerals Report. Limits of connecting reported facilities to specific products.
• TE Connectivity: 2025 Conflict Minerals Report. Company-wide supplier disclosures and product-level uncertainty.
• Microchip Technology: 2024 Conflict Minerals Report. Barriers to upstream information and working through contract manufacturers.

Related reading from Supply Chain Management Review

• Rosemary Coates. Caught between a rock and a hard place: Mapping your supply chain.
• Craig Moss. Supply Chain Due Diligence: New Regulations and What You Need to Do.

Executive takeaways

Supply chain leaders can adjust routes, suppliers and inventory when the rules are clear. The harder challenge is making those decisions when each new trade announcement changes the assumptions behind the plan.

Gene Seroka, executive director of the Port of Los Angeles, described that pressure during a keynote panel at the recent ASCM CHAINge conference in Long Beach, California. Companies are being asked to pivot repeatedly, he said, while their planning horizon becomes increasingly narrow.

“Just give us the rules. We’ll get on the ground and we’ll make things happen,” he said.

Seroka joined former U.S. Ambassador to China Nicholas Burns, Darnel Group CEO Salvador Miñarro and Tosca Derrick, a licensed customs broker and a leader in Baker Tilly’s global trade management practice. Moderator Nick Vyas guided a discussion that connected geopolitics with the practical work of moving goods, managing costs and choosing production locations.

The panel offered differing perspectives on trade policy, but a common operating challenge emerged: Companies need to understand the whole network behind a product, including the capabilities and relationships that may remain unchanged when its immediate source changes.

Put trade knowledge into operating decisions

Derrick described the strain that rapid tariff changes place on importers and the professionals who help them interpret the requirements. A headline can trigger immediate requests from companies trying to determine what it means across an entire supply chain.

The pace of those requests tests both available expertise and the systems used to process shipments, she said. Companies that previously gave duties relatively little attention now need a much deeper understanding of their exposure.

Derrick also challenged the assumption that buying domestically insulates a business from tariffs. A domestic vendor may import the product or rely on imported goods, passing some of the cost through the commercial relationship.

“I think it really depends on the business and how they set up their business,” she said.

Asked how much tariff cost reaches consumers, she declined to offer a single percentage. Margins and the structure of the business influence what it can absorb and what it passes along, she noted.

That perspective puts the supplier network inside the cost discussion. The location of the company issuing the invoice does not necessarily reveal where a product was made or which costs affected it along the way. Derrick urged greater attention to where goods come from and what is required to bring them to market.

For executives, the practical implication is to connect trade knowledge with purchasing, cost analysis and customer decisions, rather than treating it only as a shipment-clearance concern, she noted.

Examine what changes when sourcing moves

Seroka described a shift in the Port of Los Angeles’ business across Asia. He said China’s share of the port’s business portfolio had fallen over the past decade while imports from Vietnam through Los Angeles had grown substantially.

The shift creates opportunities, but he cautioned that rapid growth tests the receiving country’s ports, inland transportation, manufacturing capacity, workforce and training systems.

Changing the country that ships a product does not necessarily mean replacing every capability behind it. Seroka said Chinese investment is part of the growth in manufacturing elsewhere in Asia, alongside new relationships and companies’ efforts to diversify sourcing.

Vyas raised the issue of suppliers at multiple tiers and the challenge of understanding the production network beyond the immediate source. The discussion pointed toward a broader assessment of what a sourcing change accomplishes: Which activities move, which dependencies remain and what capabilities the alternative location can support.

For companies making those decisions, the operating ecosystem belongs in the business case. A location that offers an attractive cost or route still needs the infrastructure and supplier support required to deliver consistently.

Evaluate regional opportunities with their constraints

Miñarro sees the changing trade environment as an opportunity for Latin American manufacturing. He described Darnel’s combination of production and distribution across the region and said volatility can change both demand and the way companies serve it.

“I see the current volatility as opportunity, especially for Latin America.”

He also emphasized that the opportunity comes with limits. Some countries are not ready to absorb additional demand, and access to electricity, natural gas and other infrastructure can affect whether a location is viable.

Miñarro argued that Mexico is an important part of the regional response but cannot be the entire answer. He sees potential in other Latin American markets, while recognizing that their readiness and policy conditions vary.

His own operating approach includes serving a local market wherever the company manufactures. Depending entirely on exports, he said, leaves profits exposed to short-term changes in cross-border trade.

That makes customer demand part of the location decision alongside production cost. A facility with access to local buyers may have options that an operation built exclusively around exports does not.

Miñarro also questioned how much production the United States can bring home quickly. In his view, workforce availability and the time needed to establish raw-material supply make partnerships with other countries an important part of reducing dependence on China.

Treat partnerships as part of competitiveness

Burns approached the discussion through the relationship between the United States and China. He argued that competition with China needs to be paired with effective relationships with other trading partners.

He expressed concern that the United States is weakening ties with countries that could be important partners, including Canada, Mexico and several Asian and European economies. His argument was that economic competitiveness depends in part on maintaining those relationships, rather than focusing solely on barriers to trade.

“We’ve got Mexico and Canada. It’s a power base here in North America,” he said.

Burns also called for a more attractive U.S. trade strategy in Asia. He described countries that have substantial economic relationships with China while also seeking security relationships with the United States.

Miñarro similarly urged the United States to invest energy in collaboration and in the conditions that would support production elsewhere. Seroka said commercial partners continue to look for ways to sustain business despite the political noise.

Keep routes and relationships in the same plan

Seroka’s logistics perspective added another layer to the sourcing discussion. He described how disruptions to major maritime routes can change transit times, fuel requirements and the ports through which cargo moves.

Those changes affect the economics of a sourcing choice even when the product and supplier remain the same. A network review therefore needs to consider how the shipment reaches its destination, alongside where production takes place.

The panel also returned repeatedly to relationships. Derrick’s account showed the importance of access to trade expertise when requirements change quickly. Miñarro emphasized collaboration across countries. Seroka described supply chain professionals relying on one another to interpret events and keep business moving.

“It’s a much bigger picture, but it’s a relationship-based business,” Seroka said.

Executive takeaways

Dealmaking is regaining momentum. Recent industry outlooks project continued growth in corporate M&A volume through 2026, with deal value expected to outpace volume as larger, more strategic transactions dominate [1].

At the same time, nearly half of CEOs surveyed1 say they plan divestitures to release capital and reduce portfolio complexity [1]. But the environment underneath this activity has changed. Capital costs remain elevated [2], tariff and trade-policy volatility continues to reshape sourcing economics [3], and roughly three-quarters of executives report that geopolitical and economic crosscurrents have altered their growth strategies [1].

The practical consequence is that deals can no longer be underwritten on multiple expansion or cheap financing. They must pay for themselves through operational value creation. And one of the most reliable operational levers sits, often underused, inside the supply chain organization: procurement.

Why procurement, and why now

For most industrial, consumer, life sciences, and technology-hardware companies, third-party spend is the single largest cost line—frequently 40% to 70% of revenue [4]. That arithmetic makes external spend the largest addressable pool in most cost-synergy cases, and it explains why procurement-driven savings routinely represent an outsized share of announced synergy targets in horizontal deals [5][6].

Procurement synergies also tend to be among the fastest to capture. Unlike footprint rationalization or systems consolidation, which can take years and significant capital, harmonizing contracts, consolidating volumes, and aligning payment terms can begin delivering results within the first 12 to 18 months after close—often funding the broader integration program [5][6]. Figure 1 shows the “Typical Value Potential” by “Time to Capture” matrix.

Figure 1

 

Yet in many transactions, procurement is brought in only after the deal is announced, handed a top-down synergy target it never had the opportunity to validate. The gap between announced and realized synergies in such deals is rarely a surprise to the category managers who inherit them.

Before the ink dries: procurement in diligence

The most consequential shift in leading practice is moving procurement upstream, into pre-sign diligence. With clean-room protocols, deal teams can build a combined spend cube, map supplier overlap, and analyze contract terms across both entities before signing—replacing the traditional “1% to 2% of spend” rule of thumb [7] with a bottom-up estimate grounded in category-level facts. The difference matters twice: first in valuation, where a defensible synergy case supports (or disciplines) the price, and second in execution, where wave plans built during diligence compress time-to-value after close. Figure 2 below calls out the cost of engaging procurement late in the deal life cycle.  

Figure 2

 

Diligence is no longer only about savings. In today’s trade environment, procurement due diligence doubles as risk diligence: supplier concentration, single-source dependencies, change-of-control clauses that can void favorable terms, and—increasingly—the tariff exposure of the combined supply base [3][8]. A target whose cost position depends on sourcing lanes now subject to shifting duties may carry a hidden liability that rivals the expected synergy value. Deal models that ignore this are mispricing risk.

Sign to close, and the first hundred days

The period between sign and close is where procurement value is either protected or quietly eroded. Clean teams can pre-negotiate sourcing strategies and prepare supplier communications without violating gun-jumping rules. Day-one readiness—contract continuity, purchase-order and payables processes, supplier onboarding—determines whether the combined company starts capturing value or firefighting supply disruptions.

After close, sequencing is the discipline that separates strong programs from stalled ones. Quick wins—adopting the better of two contracts, aligning payment terms, attacking tail spend—generate early, visible results [5]. Deeper levers follow: joint category strategies, specification harmonization, make-versus-buy decisions across the combined network. Throughout, the programs that endure are those where savings are tracked through finance-validated baselines into the P&L, not declared in a project tracker and never seen again.

The divestiture mirror image

Divestitures—rising sharply on the corporate agenda—present the same lever in reverse. Three procurement questions determine whether a carve-out preserves or destroys value.

First, contract separation: which supplier agreements convey with the divested business, which must be replicated, and which trigger repricing once volumes split. Second, stranded costs: the seller’s remaining spend base shrinks, but its supplier terms were priced on the old volumes; without proactive renegotiation, the remaining company inherits a structurally worse cost position [9]. Third, standalone capability: the divested entity needs functioning procurement—people, contracts, systems—from day one, and every month it lingers on transition service agreements adds cost and constrains both parties. Sellers who prepare procurement separation early consistently see shorter TSA timelines and stronger valuations, because buyers price the uncertainty they cannot see.

Three trends raising the stakes

Three forces are elevating procurement from a synergy line item to a strategic deal consideration.

Geopolitics has made supply base design part of the deal thesis. Consolidating to fewer, larger suppliers—the classic synergy play—can now conflict directly with resiliency objectives such as dual sourcing and regionalization. Leading acquirers evaluate the combined supply network for both cost and exposure, and sometimes conclude that the resilient answer, not the cheapest one, creates more enterprise value.

Artificial intelligence is compressing the analytical timeline. Spend classification and contract abstraction that once took months of manual effort can now be performed in weeks, making rigorous procurement diligence feasible even in compressed deal processes. The constraint is shifting from analytical capacity to data quality—a reason to invest in spend data hygiene before a deal, not during one [10].

Talent has become the fragile ingredient. Category expertise and supplier relationships walk out the door easily during integrations, and synergy models rarely price that risk. Retention planning for procurement talent deserves the same attention given to commercial and technical leadership.

The takeaway for supply chain leaders

The implication for senior supply chain executives is not simply to run procurement well, but to make it deal-ready: clean spend data, a current contract repository, documented category strategies, and a playbook for both integration and separation. Organizations with that muscle can move at deal speed—and their leaders earn a seat at the table before the price is set, when the decisions that determine value are actually made.

In a market where deals must be won operationally, procurement is no longer a back-office follow-on to the transaction. It is one of the few levers that can be sized before signing, mobilized at close, and measured in the P&L within the first year. Deal teams that treat it that way will keep more of the value they pay for.

References

1. EY-Parthenon, 2026 M&A Outlook
2. Federal Reserve / OECD economic outlook publications
3. E2open, Tariff Volatility and Global Trade Realignment
4. GEP, Procurement Cost Synergies in Mergers & Acquisitions
5. Efficio Consulting, M&A Procurement Synergies Guide
6. GEP Procurement Synergy Benchmarking Research
7. Procurement synergy estimation benchmark studies (Efficio)
8. Procurement due diligence literature on supplier concentration and change-of-control risk
9. Liberty Advisor Group, Stranded Costs, TSA Fees and Standalone Costs
10. AI-enabled procurement analytics studies from major procurement technology providers