Retailers have spent the past five years building resilience into their supply chains after shocks ranging from Brexit to Covid-19 and geopolitical conflict. But now, according to the latest news, a new risk could be emerging – AI services in the supply chain.

Last week, the US government’s decision to label Anthropic a “supply chain risk” triggered a legal battle that could reshape how companies view artificial intelligence suppliers across their technology stacks.

The AI firm, developer of the Claude model, has filed lawsuits challenging the designation after the United States Department of Defense banned the use of its technology in military systems and contracts. The move is widely seen as unprecedented: the designation has historically been used to block foreign technology providers, not domestic US companies.

For Anthropic, it is clear the stakes are high, with the warning that the decision could jeopardise hundreds of millions of dollars in revenue in the near term and potentially billions in the longer term.

But the implications stretch far beyond defence technology. Retail Gazette sat down with Pinsent Mason head of commercial Clare Francis to discuss if, given retailers and consumer brands increasing reliance on AI for demand forecasting, logistics optimisation and customer service, what happens if a core technology supplier suddenly becomes unavailable?

AI suppliers becoming part of the supply chain

According to Francis the dispute signals a deeper shift in how AI vendors are viewed.

“This shift in treating software vendors like critical infrastructure providers will certainly result in extra focus on AI providers in the supply chain,” she says.

Procurement teams are already beginning to rethink supplier risk models. “Companies are likely to factor this into their overall procurement strategies and become increasingly cautious about how and where it is used with enhanced supplier due diligence,” Francis adds.

In practice, that means technology providers may increasingly be assessed in the same way as physical suppliers, with scrutiny around resilience, concentration risk and the ability to switch providers if disruption occurs.

Retail’s growing reliance on AI is creating a new operational vulnerability: dependence on a small number of technology platforms.

Francis warns businesses need to plan for scenarios where an AI service disappears overnight, not necessarily due to outages or cyberattacks, but regulatory intervention.

“Companies will need to consider the implications if an AI provider becomes a single point of failure and how it will be able to continue to operate and meet any wider commitments,” she says.

Legal safeguards can help, including contractual protections such as termination rights or data portability clauses — but these are unlikely to prevent operational disruption if a service suddenly becomes unavailable.

The lesson, Francis argues, is familiar from traditional supply chains: diversification matters.

“It makes good commercial sense for retailers to have a robust technology stack built across a couple of key platforms or providers, so that critical infrastructure isn’t wholly reliant on one service.”

The architecture challenge

One practical solution is designing systems so they are not tied to a single AI provider. Francis suggests companies should consider building integrations using common standards that multiple providers can support.

“If they use standard data formats and APIs that multiple providers can support, it’s easier to swap providers,” Francis says.

That approach, sometimes described as “provider-agnostic” architecture, could become increasingly important as AI platforms proliferate.

For retailers experimenting with AI tools across warehouses, merchandising and customer service, Francis also warns against adopting technology without clear objectives. “We take a firm-first view and look for the problems or pain points that technology could help us solve, rather than following technology hype,” she says.

Retailers should start by defining the business problem before selecting the technology. That includes questions such as whether an AI system can adapt to seasonal demand patterns, or whether the business needs an off-the-shelf solution or a more bespoke platform.

Structured pilots and clear success metrics are also essential. “When it comes to measuring impact and value, start out with a concise set of measures that are easy to track,” Francis says.

Ultimately, the biggest indicator of success may be behavioural rather than technical.

“AI only creates value if people actually use it and trust it. Adoption is often the clearest indicator that technology is genuinely embedding itself into operations.”

A new question for investors

The Anthropic dispute may also change how investors evaluate retailers. Supply chain resilience has already become a core consideration after recent disruptions, and AI is increasingly part of that discussion.

“Investors will look beyond transparency and want proof of diversified, technology-agnostic AI architectures,” Francis says.

Businesses that can demonstrate robust contingency planning around technology risk, including operational fail-safes, may therefore appear more attractive investment prospects. However, another possibility is that the bigger shift, may be political.

The Pentagon’s move shows how governments may increasingly intervene directly in commercial AI supply chains, particularly where national security is concerned. “With a heightened focus on national security countries may now feel justified in applying restrictions on AI vendors,” Francis says.

The decision effectively elevates AI providers from software vendors to actors within the geopolitical risk landscape. “I would expect more scrutiny from governments as technology becomes embedded in critical supply chains,” she adds. “If a technology is critical to how industries and infrastructure operate, it becomes a matter of national interest.”

For retailers, the Anthropic dispute may feel distant from everyday operations. Yet the underlying issue, dependence on external AI platforms, is becoming central to how modern retail functions.

“Any disruption in the use of AI for inventory management or customer services would have a direct impact on customer experience and, therefore, on the retailer’s brand in the market,” Francis says.

From inventory management to customer experience tools, AI is increasingly embedded in the systems that underpin retail businesses, but, as Francis points out, if those systems fail, the repercussions could be immediate.

Click here to sign up to Retail Gazette‘s free daily email newsletter