Google, Microsoft & Amazon Inside the Pentagon — Anthropic Left Out
The line between Silicon Valley and the US military has officially disappeared.
The Pentagon announced new agreements with seven leading artificial intelligence companies to support classified workflows in military operations. SpaceX, OpenAI, Google, NVIDIA, Reflection AI, Microsoft and Amazon Web Services have all been selected to deploy their AI products in the Department of Defense's Impact Level 6 and Impact Level 7 network environments, with a specific focus on streamlining data synthesis, improving warfighter decision-making and elevating situational understanding. Oracle was later added as an eighth partner.
Three of those companies — Google, Microsoft and Amazon — are GAFAM members. This is not a footnote. It is a turning point.
What Impact Level 6 and 7 Actually Means
These are not administrative systems. Impact Level 6 covers classified national security information. Impact Level 7 is the highest classification tier in US military infrastructure — the networks where decisions about warfare, surveillance and weapons deployment are made.
The Pentagon framed the deals as part of its ongoing efforts to build an architecture that prevents AI vendor lock and ensures long-term flexibility for the Joint Force — making a diverse range of AI tools available for Defence employees. The language is careful. The reality is direct: Google, Microsoft and Amazon AI will now operate inside the US military's most sensitive networks.
The Anthropic Exception — and What It Reveals
Not included in the deal was Anthropic, which the Trump administration blacklisted after Anthropic insisted the Pentagon include certain safety guardrails for the government's use of AI in warfare. The Pentagon declared Anthropic a supply chain risk — a label only previously used for companies associated with foreign adversaries.
Until recently, Anthropic's Claude was the only AI model available in the Pentagon's classified network. Anthropic sued the Trump administration in response, and a federal judge in California blocked the government's effort to blacklist the company.
The contrast could not be sharper. Anthropic said: we will work with the military, but not without safety conditions. Google, Microsoft and Amazon said: we will work with the military for any lawful purpose. One company is now in court. Three are inside the Pentagon.
Google's Internal Contradiction
This deal lands at a particularly sensitive moment for Google. As we reported earlier this week, nearly 600 Google employees signed an open letter opposing the company's military AI contracts. Google's leadership is proceeding regardless. The broader scrutiny over tech company involvement with the US military has gained renewed attention amid questions over how AI has been used in the US-Israeli war with Iran.
Google's Gemini is now, simultaneously, a consumer product used by hundreds of millions and a tool available inside the Pentagon's most classified military networks. These two facts are not easy to reconcile.
What This Means for GAFAM
The Pentagon deals reveal something important about the strategic calculations being made inside the three GAFAM companies involved. Military contracts offer stable, long-term revenue that is largely insulated from market cycles. They offer preferential access to government infrastructure. And in the current Washington climate, they offer political protection.
The price is reputational — and potentially legal, in jurisdictions outside the United States.
The European Perspective
The EU AI Act explicitly identifies AI systems used in military and national security contexts as among the most sensitive applications requiring the strictest oversight. European regulators have no jurisdiction over the Pentagon. But they have jurisdiction over Google, Microsoft and Amazon's European operations. The question Brussels must now answer: can a company that deploys AI inside classified US military networks be trusted to comply with European AI safety standards? It is a question no one in Brussels is yet asking loudly. They should be. gafam.ai will be watching.