Anthropic PBC's ongoing legal dispute with the U.S. government over the potential federal ban of its AI models is creating financial threats for other companies. Design software maker Figma (FIG.US) has indicated that if Anthropic is formally designated a "supply chain risk," it could impair Figma's ability to sell products to government agencies. In a regulatory filing on Thursday, Figma disclosed that Anthropic's Claude is the large language model underpinning the AI features built into its products sold to federal agencies. The filing states that if the U.S. government blacklists Anthropic and Figma cannot find a suitable alternative, "sales to governmental agencies and highly regulated organizations could be adversely affected."

The U.S. Department of Defense's conflict with and fear of Anthropic escalated in February when the U.S. President indicated the government would place the company on a "supply chain risk" blacklist due to disputes over military use of its models. Anthropic subsequently sued the Department of Defense, alleging it was banned because of its opposition to how the technology might be deployed. Beyond the public commercial dispute, underlying the Defense Department's extreme measures is profound political fear within the U.S. regarding a secretive, super-powerful model codenamed "Claude Mythos" reportedly under development at Anthropic. According to disclosures from multiple Wall Street channels and intelligence sources, the "Mythos" model has demonstrated a startling and chilling aptitude in cybersecurity, capable of autonomously discovering highly complex "zero-day vulnerabilities" in large distributed software systems and generating precise exploitation tools without human intervention. This automated, top-tier hacker-level destructive capability has deeply alarmed decision-makers, including senior officials. There is widespread concern in Washington that if such a technology with autonomous cyber-weaponization potential were to lose absolute control—whether through malicious use by adversaries or logical deviation of the model itself—it could inflict catastrophic damage on the nation's public hospitals, banking clearance systems, and power grids. While the Defense Department reportedly continues to deploy the model under pressure to patch its own security gaps, this has also reinforced its resolve to use administrative means to completely remove this "uncontrollable" company from the core supply chain.

As the model provider is branded a supply chain security risk, Figma is highly likely to face a devastating short-term impact, including the loss of federal orders and potential mass cancellations or subscription terminations by existing government clients. More worryingly for leadership, this panic is spreading to high-value private sectors heavily guided by government compliance policies, such as finance, healthcare, and utilities. Finding, testing, and seamlessly replacing a large language model of comparable scale requires months of effort and immense, difficult-to-estimate R&D funds. Unable to achieve a timely technological decoupling, Figma's commercial monetization and revenue growth expectations for its core products are now under a severe cloud.

Figma's disclosure echoes similar risk warnings issued by other companies. Cybersecurity provider Tenable Holdings Inc. earlier this month informed investors it might need to rapidly replace Anthropic's products, which could lead to "significant engineering costs, service disruption, and loss of key product features." Global freight platform Freightos Ltd. stated last month that its use of Anthropic's models exposed it to "significant regulatory and political risk" due to the legal dispute with the Defense Department. Figma, a company providing software for designing websites and applications, was authorized to sell products to federal agencies in early 2025. According to the filing, the company expects government sales to grow over time, though they currently represent a low proportion of its overall revenue.

As Anthropic is thrust into the spotlight for adhering to safety principles, the brutal commercial competition in the AI arena has not paused. Its arch-rival, OpenAI, quickly seized this policy-driven opportunity. Within hours of Anthropic's exclusion from procurement lists, OpenAI engaged with the Defense Department and, by late February 2026, swiftly signed a strategic procurement agreement to introduce advanced AI models into military confidential networks, rapidly capturing market share. Although the initial agreement contained broad terms allowing use for "all lawful purposes," sparking internal controversy, OpenAI urgently amended the contract in early March to clearly establish three safety red lines: "prohibition of use for domestic mass surveillance, prohibition of independently guiding autonomous weapons, and prohibition of automating high-risk decision-making." Furthermore, this market vacuum was not ultimately filled by OpenAI alone; in May, the U.S. Department of Defense included eight tech giants, including OpenAI, Google, Nvidia, and xAI, as suppliers for the confidential network.

Meanwhile, the alliance between Figma and Anthropic has recently developed subtle and dangerous cracks. Notably, Anthropic's Chief Product Officer, who had long held a key position on Figma's board, officially resigned in April. Shortly thereafter, Anthropic prominently launched "Claude Design," a native product directly targeting the core of the design industry. This move has made capital markets acutely aware that SaaS software companies must not only bear the political and compliance risks from their foundational model providers but also constantly guard against these industry giants, who control core underlying technology, from crossing over to become their most devastating direct competitors.