Pentagon Signs Classified AI Deals With OpenAI, Google, Nvidia and More — Anthropic Left Out – AAXYZ

Pentagon Signs Classified AI Deals With OpenAI, Google, Nvidia and More — Anthropic Left Out

In a landmark move that reshapes the relationship between Silicon Valley and the U.S. military, the Pentagon has struck classified artificial intelligence agreements with seven major technology companies, clearing the way for their AI systems to be deployed on the Department of War’s most sensitive networks.

The deals, announced on May 1, 2026, include partnerships with OpenAI, Google, Nvidia, Microsoft, Amazon Web Services, SpaceX, and Oracle — a consortium spanning the full stack of AI infrastructure from chip design to large language models to cloud computing and satellite communications.

One company was conspicuously absent: Anthropic, the maker of Claude, which has been labeled a “supply chain risk” by Pentagon officials — a designation that has sparked both a legal challenge from Anthropic itself and fierce internal debate at Google, whose employees have publicly objected to their company’s expanded military AI work.

What the Agreements Cover

Under the new agreements, the participating tech giants will be authorized to deploy their AI technologies on the Pentagon’s classified networks — including Secret and Top Secret/SCI (Sensitive Compartmented Information) systems. This is a significant step beyond previous arrangements, which largely limited commercial AI access to unclassified military networks.

According to the Department of War’s official announcement, the AI systems will be used “for lawful operational use” across warfighting systems. The stated goal, as one source described it, is to build an “AI-first fighting force” — a fundamental transformation of how the U.S. military processes intelligence, plans operations, and makes decisions in contested environments.

The companies cleared for classified work include:

OpenAI — providing advanced language models for intelligence analysis and decision support
Google — offering its cloud AI and machine learning infrastructure, expanding beyond its existing Project Maven involvement
Nvidia — supplying GPU computing hardware and AI frameworks essential for running large models in classified environments
Microsoft — delivering Azure Government cloud services and Copilot-based AI tools
Amazon Web Services (AWS) — providing classified cloud infrastructure and AI services through AWS GovCloud
SpaceX — contributing satellite communications and potentially Starshield technology for secure data links
Oracle — offering its cloud infrastructure and database capabilities for classified workloads

Breaking Defense reported that a total of eight tech firms were cleared, with the eighth being a company called Reflection — a less well-known player in the defense AI space.

Why Anthropic Was Excluded

The most striking aspect of the announcement is the deliberate exclusion of Anthropic. According to multiple reports from Reuters, CNN, and WION, the Pentagon has formally designated Anthropic as a “supply chain risk” — a serious classification that effectively bars the company from working on the nation’s most sensitive defense AI programs.

The exclusion stems from Anthropic’s own ethical stance. The company reportedly refused to agree to a clause that would require it to remove certain safety guardrails from its models when used by the military — specifically, the restrictions that prevent AI from participating in lethal targeting or autonomous weapons systems. Sources from Republic World and The Information indicate that Anthropic is actively suing the Pentagon over this ethics clause, arguing that the requirement to waive safety constraints violates the company’s core principles and its public commitments.

This puts Anthropic in a unique position among AI companies: it is the only major U.S. AI lab to draw a hard line on military use of its technology, even at the cost of losing lucrative defense contracts. By contrast, OpenAI, Google, and Microsoft have all indicated their willingness to work within the Pentagon’s framework.

Internal Resistance at Google

The deals have not been universally welcomed even within the participating companies. Multiple sources, including reporting from The Times of India, note that Google employees have objected to the company’s expanded Pentagon AI access. This echoes Google’s internal controversy from 2018 over Project Maven, when thousands of employees signed a petition demanding the company stop working on AI for drone targeting — leading Google to eventually decline to renew the contract.

This time, however, Google appears to be moving forward despite internal objections, suggesting that the balance of power within the company may have shifted, or that the Trump administration’s defense priorities have created new pressure on tech firms to cooperate.

The Broader Context: AI and Warfare

These agreements represent the most significant institutional commitment to military AI in U.S. history. The implications extend far beyond procurement:

The Pentagon’s push for an “AI-first fighting force” signals a fundamental transformation in how the U.S. military will operate — one where AI systems play central roles in intelligence analysis, target identification, logistics planning, and potentially battlefield decision-making.

Several key dimensions are at play:

1. Intelligence Acceleration

Large language models deployed on classified networks could dramatically speed up the analysis of intercepted communications, satellite imagery, and signals intelligence. Analysts who once spent hours or days reviewing data could receive AI-generated summaries and threat assessments in minutes.

2. Decision Superiority

The U.S. military’s concept of “decision superiority” — out-thinking and out-deciding adversaries — hinges on processing information faster than opponents. AI systems running on classified networks would give commanders real-time analytical capabilities that were previously impossible.

3. Supply Chain and Sovereignty Concerns

By excluding Anthropic as a “supply chain risk” while simultaneously deepening partnerships with other firms, the Pentagon is making a statement about which companies it considers trustworthy enough to handle classified military data. This creates a two-tier AI industry in the defense sector.

4. The China Factor

While not explicitly stated in the Pentagon’s announcement, these deals are widely understood as a response to China’s own aggressive military AI development. The PLA has been integrating AI into its command-and-control systems, and the U.S. is racing to maintain technological parity or advantage.

What This Means for the AI Industry

The Pentagon deals create a new dynamic in the AI industry that will likely play out over several years:

Revenue impact: Defense contracts represent a significant revenue opportunity. For companies like OpenAI, which faces enormous infrastructure costs, government contracts could provide crucial financial support.
Regulatory advantage: Companies that work closely with the Pentagon may gain preferential access to government AI procurement, creating a moat against competitors.
Ethical divide: Anthropic’s refusal to participate and subsequent lawsuit could galvanize a movement of AI companies and researchers who believe in maintaining strict ethical boundaries around military applications.
Talent competition: Defense AI work could attract some researchers while repelling others, potentially reshaping the talent landscape at major AI labs.

What Happens Next

Several developments are expected in the coming weeks and months:

Anthropic’s lawsuit against the Pentagon over the ethics clause will likely become a focal point for debate about AI companies’ obligations — or lack thereof — to military customers. The outcome could set a precedent for how AI developers handle government defense contracts.

Implementation timeline: While the agreements have been signed, actual deployment on classified networks will take time. Security certifications, infrastructure build-out, and model hardening are all prerequisites before these systems can handle classified data.

Congressional scrutiny: Given the sensitivity of military AI, expect hearings and oversight requests from both the House and Senate Armed Services Committees. Lawmakers from both parties have expressed interest in how AI is being integrated into defense operations.

International response: NATO allies and partner nations will be watching closely. The U.S. moves in military AI typically influence allied procurement and strategy, potentially triggering similar deals between European governments and AI companies.

The Bottom Line

The Pentagon’s classified AI deals with OpenAI, Google, Nvidia, Microsoft, AWS, SpaceX, and Oracle mark a watershed moment in the convergence of artificial intelligence and military power. For the first time, the full range of commercial AI technology — from foundation models to GPU clusters to satellite communications — will be integrated into America’s most classified warfighting systems.

Anthropic’s exclusion and lawsuit add a dramatic ethical dimension to the story, forcing a public debate about where the line should be drawn between AI innovation and military application. As Google’s internal resistance demonstrates, this is not just a question for policymakers — it’s a question that tech workers, researchers, and citizens will grapple with for years to come.

The era of the “AI-first fighting force” has begun. What that force looks like — and what constraints it operates under — will be one of the defining technology-policy questions of the decade.

Sources: U.S. Department of War (war.gov), Reuters, The New York Times, The Guardian, CNN, The Wall Street Journal, The Hill, Decrypt, Breaking Defense, TechCrunch, Engadget, The Information, WION, Republic World, South China Morning Post, and others.

Pentagon Signs Classified AI Deals With OpenAI, Google, Nvidia and More — Anthropic Left Out