Every business that touches an international supply chain has spent the last eighteen months absorbing something it didn’t budget for. Not all at once, and not evenly — but absorbing it nonetheless, usually quietly, inside the cost of goods sold, hoping the next quarter would bring stability instead of another rate change.
That quiet absorption is about to become a lot less quiet. According to 2026 research from Tradeverifyd, 73% of supply chain leaders expect to hit their “tariff absorption wall” by the end of this year — the point at which costs can no longer be managed on the balance sheet and have to move onto the price tag.
Right now, 83% of organisations are absorbing at least a portion of tariff costs internally, and 46% are absorbing nearly all of them to keep prices stable for customers. Only 12% are currently passing the majority of the cost straight through. That ratio is not going to hold.
At the same time, a genuinely useful response is emerging, and it’s worth separating from the noise around it: 77% of procurement and supply leaders are now rolling out AI tools specifically to navigate tariff and geopolitical risk, according to Ivalua’s 2026 research.
This issue is about that response — not AI as a cost-cutting layer bolted onto an existing supply chain, but AI as the compliance and scenario-planning infrastructure that lets a business see a tariff coming before it lands, and prove it classified and priced correctly after it does.
Why the wall is real, not rhetorical.
The data on how businesses got here is unambiguous. Thomson Reuters’ 2026 Global Trade Report, surveying 225 senior trade professionals, found 72% now name U.S. tariff volatility as the single most impactful regulatory change of the year — up sharply from 41% just twelve months earlier.
Three-quarters (76%) now believe the current tariff posture represents a permanent shift in U.S. trade policy that will persist for at least the next four years, not a temporary disruption to wait out. And the share of companies now absorbing or considering absorbing tariff costs internally rather than passing them on has tripled, from 13% to 39%, in a single year.
McKinsey’s own December 2025 survey of supply chain leaders found 82% report their supply chains have been directly affected by new tariffs, with increased material costs and softer customer demand both showing up as consequences.
This isn’t a story confined to multinationals, either — Netstock’s 2026 SMB research found 73% of small and mid-sized businesses have pushed their inventory planning further out specifically because of tariff uncertainty, and more than half now say the tariff impact on their business is worse than it was a year ago.
The honest reading of this data: the wall isn’t a single event. It’s the point at which a strategy of quiet absorption stops being financially available to most businesses, roughly on the same timeline, regardless of size.
Beyond efficiency: what AI is actually being asked to do here.
It’s tempting to file “AI in supply chain” under the same efficiency story as everything else — faster processing, fewer people, lower cost per transaction. That’s not what’s driving the 77% adoption figure. The two jobs AI is actually being recruited for in this environment are more specific and more directly tied to the tariff problem itself.
Classification, done defensibly. Getting a product’s tariff classification wrong is expensive in both directions — classify too high, and you overpay duty on every shipment; classify too low, and you carry real penalty and audit exposure.
It’s also a genuinely hard problem: the U.S. Harmonised Tariff Schedule alone runs to more than 17,000 line items, and in datasets where trained trade professionals independently classify the same product, they disagree with each other at least 30% of the time. That’s not a training gap — it’s inherent ambiguity in the system.
#AI classification tools, such as Thomson Reuters’ ONESOURCE Global Classification, are now targeting roughly 95% prediction accuracy against historical rulings, with early implementations reporting a 50% reduction in time spent on classification work — and, just as importantly, generating an audit trail that cites the specific tariff rule and prior ruling behind each decision, which is exactly what a business needs to demonstrate “reasonable care” if customs comes asking.
The credible tools in this space keep a human reviewer in the loop rather than positioning AI as a full replacement for trade compliance expertise.
#Scenario planning, before the tariff is finalised. The second use case is forward-looking rather than reactive: AI systems that continuously scan regulatory filings, trade announcements, and geopolitical signals across languages and jurisdictions, then map that intelligence against a company’s specific supplier and product base to flag a likely tariff change before it’s formally announced — turning trade risk management from a quarterly review into a continuous one.
This is the use case behind the sharpest data point in Ivalua’s research: 98% of U.S. businesses with fully deployed AI tools report feeling prepared for geopolitical risk, against just 11% of businesses that are still merely considering AI deployment who describe themselves as well prepared.
That’s not a marginal advantage. It’s the difference between a trade team finding out about a tariff from a Federal Register notice and one that saw it coming three weeks earlier, in time to adjust sourcing or reprice before it hit.
The adoption-versus-trust gap, here too.
If you’ve read previous issues of this newsletter, you’ll recognise the shape of what comes next: strong adoption numbers sitting alongside real caution about how far to trust the technology with consequential decisions. Supply chain isn’t an exception.
Research from Pull Logic puts the global AI-in-supply-chain market at roughly $19.8 billion in 2026, with organisations that have matured their deployments reporting an average ROI of 307% within eighteen months. Confidence is genuinely rising — 67% of supply chain leaders say they trust AI more than they did a year ago. But only 10% say they trust it enough to hand it critical, high-stakes decisions outright.
That caution looks appropriate rather than excessive, given what’s actually at stake. A misclassified HS code or a missed tariff signal doesn’t just cost money — it can trigger a customs audit or a compliance investigation, and the U.S. Department of Justice has reportedly made trade enforcement one of its top priorities this year.
The businesses getting real value from AI here are treating it the same way the better-run finance teams treat agentic AI in general: as a tool that accelerates and documents a decision a qualified person is still accountable for, not a system that gets left to run unsupervised because the sales demo looked impressive.
There’s also a structural shift worth noting alongside the technology: Thomson Reuters’ data shows trade functions are increasingly being pulled into cross-functional collaboration as a direct result of tariff pressure — most often with Finance (50% of trade teams report closer collaboration), Operations (46%), and Procurement (30%).
Where that’s happening, 43% of trade professionals report more influence over procurement decisions meaningfully, and 37% report more involvement in executive decision-making generally.
The tariff wall is quietly elevating trade compliance from a back-office function to a seat in the room where sourcing and pricing decisions get made — and AI tooling is a large part of what’s making that elevation practical.
What this means for your business.
You don’t need to be moving billions of dollars of freight across borders for this to be relevant. If you import components, materials, or finished goods from anywhere, some version of this applies to you.
Get your classification defensible before you need it to be. Whether or not you invest in dedicated AI classification software, make sure whatever you’re using today — a broker, a spreadsheet, institutional memory — produces a documented rationale for each code, not just the code itself.
That documentation is what “reasonable care” actually looks like if you’re ever audited, and it’s precisely what the AI tools in this space are built to generate as a byproduct of the classification itself.
Treat scenario monitoring as infrastructure, not a project. The preparedness gap in Ivalua’s data — 98% versus 11% — isn’t about AI sophistication. It’s about whether a business has any continuous mechanism for seeing tariff and geopolitical risk coming at all, versus finding out when it’s already law. Even a modest, well-scoped monitoring setup closes most of that gap.
Plan your pricing conversation before the wall, not after. If 73% of your peers expect to hit their absorption limit this year, the businesses that reprice thoughtfully and early, with a clear story for customers, will fare better than the ones forced into a sudden, unexplained increase once the balance sheet can’t take any more.
Keep a qualified person accountable for every AI-assisted trade decision. The data is consistent across every function we’ve covered in this newsletter: the businesses getting hurt by AI adoption aren’t the cautious ones — they’re the ones that let confidence in the tool outrun oversight of what it’s actually deciding.
The tariff wall isn’t a temporary storm to wait out; the data suggests most trade professionals now believe it’s the new baseline. The businesses treating AI as compliance and foresight infrastructure — not just a cost lever — are the ones that will meet that wall with a plan, rather than a surprise.
Sources:
- Tradeverifyd, 79 Supply Chain Statistics To Know in 2026 (proprietary survey data, 2025–2026)
- Thomson Reuters, 2026 Global Trade Report (survey of 225 senior trade professionals, February 2026)
- McKinsey & Company, supply chain leaders survey (December 2025), as cited in SupplyChainStrategy.media and Indigrowth’s Supply Chain 2026: Tariffs, Nearshoring & Resilience
- Netstock, Tariff Impact Report 2026: 5 Proactive Strategies for SMBs
- Ivalua, How Tariffs Impact Procurement and Supply Chains in 2026 and How Geopolitical Disruption is Reshaping Supply Chains (with Sapio Research)
- Thomson Reuters / Tax & Accounting, “Transform Your Trade Compliance Workflow: How AI Eliminates the Classification Guesswork” (ONESOURCE Global Classification data, April 2026)
- TariffLens, “Using AI for HTS Classification: A Practical Guide for Trade Compliance” (classification disagreement-rate data, January 2026)
- Pull Logic, “How New Tariffs Are Reshaping Supply Chain Planning in 2026”
- KPMG, The State of Next-Gen Supply Chain: A Leadership Survey (March 2026)
As always at The AI Compass: every statistic above is sourced from a named, dated report — verify anything you plan to build a decision around, and treat single-source numbers with appropriate caution.
https://www.linkedin.com/pulse/ai-meets-tariff-wall-debiprasad-bandopadhyay-ul0df
