The Department of Defense quietly flagged Anthropic — maker of the Claude family of AI models and one of the most prominent artificial intelligence companies in the world — as a potential supply chain risk. Then Microsoft stepped in to fight it. What followed is a story about geopolitics, corporate rivalry, and the tangled question of who gets to build AI for the American military.
The designation, which has not been widely reported outside of defense procurement circles, effectively threatened to freeze Anthropic out of Pentagon contracts. A supply chain risk label from the DoD is not a minor bureaucratic footnote. It can bar a company from selling products and services to the U.S. military, and it sends a chilling signal to the broader federal contracting apparatus. For a company like Anthropic, which has been aggressively pursuing government business, the implications were severe.
Microsoft, which has invested billions of dollars in OpenAI — Anthropic’s chief rival — might seem like an unlikely advocate. But as Slashdot reported, the Redmond giant moved to support Anthropic’s push to reverse the designation. The reasoning is more strategic than altruistic, and it reveals the complex interdependencies forming across the AI industry as it races to secure government contracts worth tens of billions of dollars.
At the heart of the DoD’s concern is foreign investment. Anthropic has received substantial funding from international sources, most notably a reported $2 billion commitment from Google, but also capital tied to sovereign wealth funds and foreign entities. The Pentagon’s supply chain risk framework, governed in part by Section 889 of the National Defense Authorization Act and related procurement regulations, is designed to prevent adversarial nations from gaining access to sensitive U.S. defense systems through commercial technology vendors. The question the DoD raised was straightforward: Does Anthropic’s investor base create unacceptable exposure?
Anthropic says no. The San Francisco-based company, founded in 2021 by former OpenAI executives Dario and Daniela Amodei, has structured its corporate governance to insulate operational decisions from investor influence. It operates as a public benefit corporation, a legal structure that gives its board latitude to prioritize safety and public interest over pure shareholder returns. The company has argued that no foreign investor holds controlling influence over its technology, its data, or its government-facing operations.
But the Pentagon wasn’t immediately persuaded.
The supply chain risk designation process is opaque by design. The Federal Acquisition Security Council, an interagency body, can issue exclusion and removal orders against vendors deemed risky. These proceedings aren’t public, and companies often learn about them only when contracts start getting held up or denied. Anthropic found itself in exactly this position — watching potential defense deals stall as the designation worked its way through the bureaucracy.
Enter Microsoft. The company’s Azure cloud platform is one of the primary vehicles through which AI models, including those from both OpenAI and Anthropic, are delivered to government customers. Microsoft holds the Joint Enterprise Defense Infrastructure (JEDI) contract’s successor arrangement and remains one of the Pentagon’s most critical cloud providers. If Anthropic’s models are blacklisted, Microsoft loses a selling point: the ability to offer federal clients a choice of frontier AI models on its platform.
There’s a financial logic too. Microsoft has been positioning Azure as a model-agnostic platform. It hosts OpenAI’s GPT models, but it also offers Anthropic’s Claude and models from Meta, Mistral, and others through its Azure AI services. A Pentagon ban on Anthropic would undercut that positioning and potentially push defense customers toward Google Cloud, which has its own deep ties to Anthropic and its own federal cloud certifications.
So Microsoft lobbied. According to people familiar with the matter, Microsoft engaged with senior DoD procurement officials and members of the Federal Acquisition Security Council to argue that Anthropic’s foreign investment structure did not meet the threshold for a supply chain risk designation. Microsoft’s argument leaned heavily on the precedent it would set. If passive foreign investment — even at scale — triggered supply chain exclusions, a significant portion of the American tech industry could find itself locked out of defense work. Venture capital is global. So is the capital flowing into AI.
This is not an abstract concern. Amazon has invested up to $4 billion in Anthropic. Google has committed $2 billion. Saudi Arabia’s sovereign wealth fund and other Middle Eastern investors have poured money into AI companies across the board. If the Pentagon draws the line at any meaningful foreign capital exposure, the pool of eligible AI vendors shrinks dramatically — possibly to just the hyperscalers themselves and a handful of defense-native contractors.
That outcome would please some in Washington. Hawkish members of Congress have been pushing for stricter controls on foreign influence in AI development, particularly as it relates to defense and intelligence applications. Senator Mark Warner and others on the Senate Intelligence Committee have raised concerns about the national security implications of foreign capital in American AI firms. Their worry isn’t necessarily about Google or Amazon. It’s about the less visible investors — the limited partners in venture funds, the sovereign wealth vehicles, the layered financial structures that can obscure ultimate beneficial ownership.
Anthropic’s case is complicated by its sheer prominence. Claude has become one of the two or three most capable large language models in the world, and Anthropic has been explicit about wanting to work with the U.S. government on national security applications. The company opened a Washington office, hired former defense and intelligence officials, and began pursuing FedRAMP authorization for its models. Losing access to defense contracts wouldn’t just be a revenue hit — it would be a strategic blow at a moment when the company is trying to establish itself as a trusted government partner.
The timing matters. The AI arms race between the United States and China has become a central preoccupation of the national security establishment. The Pentagon’s Chief Digital and AI Office, led by Craig Martell until his recent departure, has been pushing to accelerate AI adoption across the military. The Replicator initiative aims to deploy autonomous systems at scale. Every branch of the armed forces is experimenting with large language models for intelligence analysis, logistics optimization, and operational planning. Cutting off one of the world’s leading AI companies from this work because of its cap table would strike many defense technologists as counterproductive.
And yet the risk is real. The U.S. intelligence community has warned repeatedly about the threat of foreign adversaries using commercial investment as a vector for espionage and influence. China’s national intelligence law compels Chinese companies and citizens to cooperate with state intelligence work. Russian and Chinese venture investments in Silicon Valley have drawn scrutiny from the Committee on Foreign Investment in the United States (CFIUS) for years. The Pentagon’s caution isn’t paranoia. It’s institutional memory.
Microsoft’s intervention on Anthropic’s behalf also raises questions about market dynamics. Microsoft and Anthropic are not natural allies. Microsoft’s multibillion-dollar partnership with OpenAI is the most consequential corporate relationship in AI. OpenAI and Anthropic compete fiercely for talent, customers, and influence. By backing Anthropic in this fight, Microsoft is playing a longer game — preserving the competitive environment on its own platform and ensuring that its government cloud business isn’t constrained by a rival’s exclusion.
There’s also a defensive angle. If the Pentagon can designate Anthropic a supply chain risk based on foreign investment, it could theoretically apply the same logic to OpenAI, which has taken money from investors with international ties, or to Microsoft itself, which generates enormous revenue from foreign governments and corporations. Precedent matters in procurement law. Microsoft’s legal and government affairs teams understand this well.
The outcome remains uncertain. As of this writing, the supply chain risk designation has not been formally finalized or withdrawn. Anthropic continues to engage with the DoD through formal channels, and Microsoft’s advocacy has reportedly given some Pentagon officials pause. But the process is slow, and the interagency dynamics are complex. The Federal Acquisition Security Council includes representatives from the Department of Homeland Security, the Office of the Director of National Intelligence, and the Department of Defense — each with its own threat calculus and institutional priorities.
What’s clear is that the fight over Anthropic’s status is a proxy for a much larger question: How does the United States balance its need for the most advanced AI capabilities with its need to control who has access to its most sensitive defense systems? The answer will shape not just Anthropic’s future, but the structure of the entire defense AI market for years to come.
The commercial stakes are staggering. Federal spending on AI is projected to exceed $15 billion annually within the next few years, according to estimates from Bloomberg Government. The defense and intelligence portions of that spending are growing fastest. Companies that secure early positioning in this market will have durable advantages — not just in revenue, but in the feedback loops that come from working with the world’s most demanding customer. Access to military use cases generates data, expertise, and credibility that feed back into commercial products.
Anthropic knows this. So does Microsoft. And so does Google, which has its own complicated history with Pentagon contracts — it famously withdrew from Project Maven in 2018 after employee protests, only to quietly rebuild its defense business in subsequent years. Google Cloud now holds significant federal certifications and actively competes for defense AI work. If Anthropic is locked out, Google’s partnership with the company becomes a liability rather than an asset in the federal market.
Amazon Web Services, Anthropic’s other major cloud partner and investor, faces similar calculations. AWS holds the other half of the Pentagon’s cloud infrastructure business and has been investing heavily in classified cloud environments. An Anthropic exclusion would ripple through AWS’s federal sales strategy as well.
The broader AI industry is watching closely. Startups seeking defense contracts are now scrutinizing their own investor bases with new urgency. Law firms specializing in government contracts report increased demand for CFIUS-related due diligence and supply chain risk assessments. The message from the Anthropic situation is clear: in the emerging AI-defense market, your cap table is a national security document.
For Anthropic, the path forward likely involves some combination of structural remedies — perhaps creating a U.S.-only subsidiary for defense work, establishing a special security agreement, or accepting restrictions on which investors have visibility into government contracts. These are well-worn tools in the defense industrial base. Foreign-owned companies like BAE Systems and Leonardo DRS have operated in the U.S. defense market for decades under similar arrangements.
But AI is different from building ships or radar systems. The models themselves are the product, and they’re trained on data that flows across organizational boundaries. Isolating a government-facing AI operation from its parent company’s research and development is technically and operationally difficult. The Pentagon will need to decide how much risk it’s willing to accept — and how much capability it’s willing to forgo — in pursuit of supply chain purity.
Microsoft’s willingness to spend political capital on Anthropic’s behalf suggests that the industry sees the current situation as untenable. A world in which the Pentagon’s AI options are limited to a handful of American-only-funded companies is a world with less competition, less innovation, and potentially worse outcomes for national security. That’s the argument Microsoft is making, and it resonates with many technologists and defense reformers.
Whether it resonates with the security hawks who ultimately control procurement policy is another matter entirely.
WebProNews is an iEntry Publication