Anthropic has experienced significant growth, driven primarily by enterprise demand for AI systems. Anthropic CEO Dario Amodei told CNBC in February that about 80% of the company’s business now comes from enterprise customers, in contrast to rival OpenAI, whose product is drawing much of its early momentum from consumer adoption of ChatGPT. The company’s annual revenue run rate is approaching $20 billion, up from about $14 billion just a few weeks ago, while a recent new round of $30 billion in funding values the AI developer at about $380 billion, the people said.
But the sudden high-stakes battle between AI startups and the Trump administration will leave customers and investors alike asking, “Will the momentum continue?”
Defense contractors have stopped providing Anthropic’s technology after the Trump administration announced last week that it would designate the company as a supply chain risk. This comes after the Department of Defense rejected the terms of its use of AI, citing safety concerns, a term previously only used for organizations believed to be controlled by foreign governments such as China or Russia when national security or espionage concerns were raised. The defense contractor’s move is not surprising. “Most of our companies are actively involved in large defense contracts, so they have very strict interpretations of the requirements,” Alexander Herstrick, managing partner at J2 Ventures, which supports startups in this space, told CNBC.
But other tech industry executives say there will be, if not already, an inevitable conversation in every corporate boardroom about human risks that goes far beyond the defense sector.
“The administration did not simply cancel the contract with Anthropic. President Trump directed federal agencies to phase out Anthropic’s technology, and the Department of Defense reportedly applied a ‘supply chain risk’ designation. These words matter,” said Spencer Penn, co-founder and CEO of LightSource, an AI-powered procurement platform.
“This situation is different.”
In a rapidly evolving world where enterprises deploy large-scale language models for AI, choosing a foundational model increasingly resembles an infrastructure decision rather than a simple software purchase, which means enterprises evaluate not just technical performance but also reputational, geopolitical, and customer perception risks, according to Penn. “The board cares about it. The risk committee cares about it. The customers absolutely care about it,” Penn said.
Anthropic did not respond to a request for comment.
Tensions between the government and Anthropic over the safety of AI and the military use of the technology have helped boost the company’s brand among consumers. On February 28, the day after the controversy, Anthropic’s Claude chatbot secured the top spot on Apple’s ranking of top US free apps, surpassing ChatGPT and placing Google’s Gemini further down the list. But as developers and large enterprises increasingly rely on AI tools to automate parts of the software development process, one of the company’s fastest-growing products is Anthropic’s coding assistant, Claude Code, which generates billions of dollars in revenue on an annual basis. This also includes tools designed to help developers create and review software and support day-to-day business operations.
Anthropic argued that the supply chain risk designation was legally unsound, advised commercial customers that it was “unaffected” and indicated it intended to challenge the decision in court. Many legal experts agree with Anthropic’s view that the government’s statement that supply chain risk designations may limit not only what private companies can do under certain government procurement and use scenarios, but also other commercial activities by private companies, goes far beyond its statutory authority.
Anthropic CEO Dario Amodei walks out during a hearing of the Senate Judiciary Subcommittee on Privacy, Technology, and the Law on Tuesday, July 25, 2023, in Washington, DC.
Bloomberg | Bloomberg | Getty Images
Anthropic has received some support within the tech industry, with trade groups representing the industry sending a letter to Defense Secretary Pete Hegseth this week expressing concerns about the designation of the U.S. company as a supply chain risk. But even though Anthropic has been a critical technology to the success of military operations in Iran, the government has so far shown little intention of softening its stance.
Anthropic’s warranty alone is not enough to satisfy many companies. “When a supplier comes in and does a good job, most teams don’t actively start looking for reasons to resume hard work. This situation is different,” Penn said.
“They closed the door. They weren’t willing to do business with us,” Pentagon Chief Technology Officer Emile Michael told CNBC’s Morgan Brennan this week. “I think their culture and their own constitution with their souls and their own values are really at odds with each other. It’s strange in a way that they want to do business with the Department of the Army, as they have done for three years, but they don’t want us to do the work of the Department of the Army. So if we end up where we end up and ultimately face it and they don’t want to do business with us, I think that’s their choice.”
Single provider risks in the race for AI adoption
Michael Murphy, partner and global AI readiness lead at Adaptivete, a consulting firm that advises large enterprises on AI implementation, said Fortune 500 procurement teams act quickly when major technology vendors face regulatory scrutiny. “Recognized compliance risks can cascade through their respective regulatory obligations,” he said, noting that the situation could force a broader shift away from reliance on a single AI provider, which is already underway within many organizations.
The government said its battle with Anthropic and its controversial award of a new contract to OpenAI last week were in part to address single-vendor concentration. “We can’t rely on one more provider. That’s what was going on even before I took on this role, and that has to change,” Pentagon Chief Technology Officer Michael told CNBC.
This will become a problem for many companies in the future.
“Over-reliance on one AI vendor is increasingly seen as a risk,” Murphy said. “Many companies are already evaluating multiple providers simultaneously to create redundancy in their AI stack.”
“More mature companies understand that each vendor plays a different role in a larger puzzle. There is power in the ecosystem, but there is also lock-in risk,” said Joshua Morley, global head of AI, data and analytics at Accodis, the technology consulting division of Adecco Group.
Ultimately, political and legal battles could accelerate an already ongoing process in which corporate decision makers diversify their AI bets among companies in the space after initial experimentation with a single vendor. Disney Chief Financial Officer Hugh Johnston recently told CNBC that the company’s initial efforts were with OpenAI, but he expects the scope to expand in the future. “We’re very open about this,” he said on CNBC’s “Squawk Box.” There will be a period of time where we’ll only use OpenAI, but it’s a relatively short period of time. We need people to actually play with the models. I’d be surprised if we don’t have multiple models instead of a single model in the future.
“This looks like a short-term disruption rather than a structural change,” Penn said. “As enterprises continue to deploy AI capabilities, they are likely to move towards a more diverse ecosystem rather than relying on a single provider.”
Supply chain risk management classifications can have a significant impact on contractors and subcontractors that rely on the technology, prompting companies to review contracts, delay implementation, and evaluate alternative AI vendors. If this designation appears to be permanent, especially for companies with dual-use across commercial and defense markets, Penn said he expects a quiet evaluation of alternative foundation model providers. “Not necessarily because teams want to switch, but because concentration risk and qualifying risk are what serious procurement organizations are paid to manage,” he said. “Most companies aren’t going to make architecture changes in a matter of days, but they’re going to start reviewing them right away. Legal departments will assess what the directive actually requires. Compliance departments will assess the risks. Security departments will ask about contingency plans,” he added.
For Anthropic investors such as Amazon, Microsoft, Nvidia and sovereign wealth funds around the world, the dispute could hinder Anthropic’s rapid expansion. “Any time a government takes aggressive action against a technology company, there is a risk,” said Brad Harrison, founder of Scout Ventures, an early-stage venture capital firm that invests at the intersection of national security and critical technology innovation. “And the worst thing you can do when you have a lot of momentum is that you’re exposed to big risks that require your time and attention,” he said.
Ben, co-founder and general partner of A16Z, an investor in Anthropic competitors OpenAI and xAI Horowitz told CNBC this week from a defense technology conference: “Just a week ago, Anthropic was accusing a Chinese company of stealing all the intellectual property from its models. Do you think the Chinese government is being restricted by DeepSeek in how it can use Anthropic technology? So we are very sympathetic to A16Z’s position.” Department of War on this matter. ”
As with many things in this administration, policy signals can change rapidly. “Constructive conversations between President Trump and Dario Amodei could lead to a softening or even more hardening,” Penn said. Federal Communications Commission Chairman Brendan Carr told CNBC this week that Anthropic “made a mistake” in its dealings with the Pentagon and “should try to correct course as best we can.”
At least for now, the unusually public nature of the conflict could accelerate debate about risk. “Typically, these types of eligibility issues proceed quietly through legal channels,” Penn said. “In this case, it became front page news.”
