The Pentagon's $200 Billion AI Shakeup: How 7 Tech G...

In one of the most dramatic shifts in American tech history, the Pentagon just handed nearly $200 billion in AI contracts to seven companies—and told Anthropic to get lost.

The Department of Defense announced Friday it has reached formal agreements with SpaceX, OpenAI, Google, Microsoft, Amazon Web Services, Nvidia, and Reflection AI to deploy their artificial intelligence systems on classified military networks. The deals are worth up to $200 billion combined, according to defense officials, and mark the largest AI procurement in American history.

Notably absent from the announcement: Anthropic, the maker of Claude, which the Trump administration blacklisted earlier this year.

This isn't just a contract dispute. It's a fundamental restructuring of who controls the most powerful AI systems in the world—and what they can be used for.

The Deal That Changed Everything

Until February, Anthropic held a privileged position in America's defense AI infrastructure. Claude was the only frontier AI model deployed on the Pentagon's classified networks, handling sensitive planning, logistics, and targeting workflows for 1.3 million Department of Defense personnel through the GenAI.mil platform.

Then everything fell apart.

The breaking point came when the Pentagon demanded that Anthropic allow its AI models to be used for "all lawful purposes"—language that would have permitted deployment in fully autonomous weapons systems and mass domestic surveillance programs. Anthropic refused, insisting on contractual guarantees that its technology would never be used to kill without human oversight or to spy on Americans.

Defense Secretary Pete Hegseth's response was swift and brutal. In March, he designated Anthropic a "supply chain risk to national security"—a designation typically reserved for companies from adversarial nations like China and Russia. The move effectively banned any federal agency, defense contractor, or government supplier from using Anthropic's technology.

Anthropic sued. In April, a federal judge in California temporarily blocked the government's effort. But the damage was done. The Pentagon had already moved on.

The Seven Horsemen of AI Defense

The Pentagon's new roster reads like a who's who of American tech—and a careful calibration of competing interests.

SpaceX and xAI bring Elon Musk's unique position as a defense contractor with proven launch capabilities. Despite his frequent criticism of government spending, Musk's companies have deep Pentagon relationships through SpaceX's Starlink and national security payloads.

OpenAI gets its first major defense contract, moving in to fill the gap left by Anthropic. The company amended its original contract in March after public backlash, explicitly prohibiting domestic surveillance of U.S. persons and excluding intelligence agencies like the NSA from using its services under the deal.

Google, which the Pentagon already uses through existing infrastructure contracts, has signed a deal enabling broader deployment of its AI models for classified work—one of the most significant developments in the company's troubled history with military AI.

Microsoft and Amazon Web Services already have deep defense relationships. The new agreements consolidate their positions as the dominant cloud infrastructure providers for government workloads.

Nvidia gets its first formal defense AI deal, a remarkable development given that the chipmaker's powerful GPUs currently power essentially all frontier AI systems. The company has agreed not to impose usage restrictions beyond what U.S. law requires.

Reflection AI, the least-known company in the group, raised $2 billion last year and is backed by 1789 Capital, a venture firm in which Donald Trump Jr. is a partner and investor.

The $200 Billion Question

The dollar figure attached to these agreements is staggering—and raises immediate questions about what exactly the Pentagon is buying.

Defense Department Chief Technology Officer Emil Michael told CNBC on Friday that the deals would allow the military to deploy AI systems "for lawful operational use" across its most sensitive networks. The systems will be integrated into Impact Level 6 and IL7 environments—the highly classified networks used for secret and top-secret work.

"This approach allows the department to continuously adopt the best AI innovations available while maintaining control over their data, architecture, and long-term technology direction," Oracle said in a statement, reflecting the Pentagon's emphasis on avoiding "vendor lock."

But the sheer scale of the commitments has caught the attention of defense analysts and Washington observers alike.

"You have to ask what you're actually getting for $200 billion," said one defense contractor who spoke on condition of anonymity. "These are framework agreements. The actual task orders will determine whether this is a genuine transformation or just another defense procurement boondoggle."

The agreements follow the Pentagon's mandate to create an "AI-first war department," with AI tools expected to streamline data synthesis, elevate situational understanding, and augment decision-making in complex operational environments.

The Anthropic Question

Even as the Pentagon celebrates its new partnerships, one question hangs over the entire announcement: What happens to Anthropic?

President Trump told CNBC last week that a deal with the company was "possible," adding that "they're very smart, and I think they can be of great use. I like smart people."

In recent weeks, Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles at the White House following Anthropic's announcement of Mythos, a new AI model with advanced cybersecurity capabilities that can simultaneously identify vulnerabilities and present a roadmap for exploiting them—a capability that one official described as a "separate national security moment."

Defense officials have noted that while Anthropic remains designated as a supply-chain risk, Mythos represents a different calculation. The tool's ability to penetrate and patch cybersecurity systems may be too valuable to leave on the shelf, regardless of the company's broader dispute with the Pentagon.

For now, the door appears slightly ajar. But Anthropic's path back to defense contracts will require significant concessions that the company has thus far refused to make—or a resolution of the ongoing litigation that could take years.

What This Means

Friday's announcement represents the most significant restructuring of American defense AI infrastructure since the creation of the internet itself. The Pentagon has explicitly chosen to expand its vendor base, avoid dependence on any single provider, and embrace a spectrum of AI capabilities—from Google's enterprise tools to OpenAI's frontier models to Nvidia's chip infrastructure.

The message from Washington is clear: AI safety concerns, while valid, will not be permitted to stand in the way of American military dominance. The Department of Defense wants every available tool, and it wants them now.

Whether this accelerates the militarization of artificial intelligence in ways that endanger civil liberties—or positions America to maintain its technological edge over adversaries—will be determined in the years ahead.

One thing is certain: The era of depending on a single AI provider for America's most sensitive defense work is over.