The Pentagon announced Friday that it has reached agreements with seven leading AI companies to deploy their technology for classified military work: SpaceX, OpenAI, Google, Nvidia, Reflection AI, Microsoft, and Amazon Web Services. The notable absence: Anthropic, maker of the Claude chatbot and the military's previously preferred AI model.

"These agreements accelerate the transformation toward establishing the United States military as an AI-first fighting force," the Pentagon said in a statement, adding that the deals "will strengthen our warfighters' ability to maintain decision superiority across all domains of warfare."

The critical detail: all seven companies agreed to the U.S. military's deployment of their technology for "any lawful use." Anthropic rejected that standard in a high-profile feud with the Defense Department last month, insisting on guardrails around potential military applications of its AI models.

$54 Billion and an AI Arms Race

The deals come as the Pentagon has requested $54 billion for the development of autonomous weapons alone. The Department of Defense is budgeting tens of billions more for AI programs related to intelligence, drone warfare, and classified and unclassified information networks. How each company's technology will be deployed was not specified.

One of the signed companies, Reflection AI, has not even released a publicly available model yet. The two-year-old startup's inclusion alongside established players like Google and Nvidia signals the Pentagon's appetite for securing AI access broadly rather than betting on proven vendors only.

Anthropic's Costly Principle

Anthropic's exclusion is not just symbolic. According to CNBC, the Pentagon's tech chief has confirmed the company remains blacklisted as a supply chain risk. Defense contractors must now certify that they do not use Anthropic's Claude models in their military work. The Pentagon has ordered existing Anthropic tools removed over the next six months.

But Reuters reports that Pentagon staffers, former officials, and IT contractors who work closely with the military are reluctant to give up Claude, which they view as superior to the alternatives. The gap between what the military wants to use and what the military is allowed to use has become a real operational friction point.

The timing creates an interesting paradox for Anthropic. The company is riding a historic commercial wave. Polymarket traders are pricing an 87% implied probability that Anthropic will surpass OpenAI's valuation in 2026, with secondary market shares trading above $1 trillion since mid-April. But the world's largest single buyer of technology just locked in deals with every major competitor and formally excluded Anthropic from the table.

The "Any Lawful Use" Standard

The phrase "any lawful use" is doing a lot of work in these agreements. It means the Pentagon can deploy these AI systems for autonomous weapons development, targeting, surveillance, cyber operations, or any other military application that falls within U.S. law. The companies have ceded control over how their technology is used once it enters the classified domain.

OpenAI's inclusion is particularly notable. The company that once published a charter promising to ensure AI "benefits all of humanity" has now agreed to unrestricted military deployment of its technology. Google, which faced employee revolts over Project Maven in 2018, has also signed on without visible resistance.

Anthropic is the only major AI company that drew a line and paid a price for it. Whether that is principled leadership or a strategic miscalculation depends on how the next two years play out. What is not debatable: the U.S. military just made its AI alliances official, and the most capable model according to its own users is on the outside looking in.