Pentagon Signs Classified AI Deals with Seven Tech Giants — Leaving Anthropic on the Sidelines – AAXYZ

Pentagon Signs Classified AI Deals with Seven Tech Giants — Leaving Anthropic on the Sidelines

On Friday, May 1, 2026, the U.S. Department of Defense announced a sweeping set of artificial intelligence agreements with seven major technology companies: Nvidia, Microsoft, Amazon Web Services, SpaceX, OpenAI, Google, and Reflection AI. The deals authorize these firms to deploy their AI systems on the Pentagon’s most sensitive classified networks, marking a significant expansion of the military’s AI infrastructure.

One company was conspicuously absent from the list: Anthropic, the AI lab behind Claude, which the Pentagon had previously labeled a supply-chain security risk.

What the Deals Cover

The newly signed agreements grant access to Impact Level 6 and 7 network environments — the two highest classification tiers used by the U.S. military. These environments handle the most sensitive defense operations, including intelligence analysis, mission planning, and targeting systems.

Amazon Web Services finalized its agreement late Thursday, just hours before the Pentagon’s Friday announcement, according to two Pentagon officials briefed on the matter. Friday’s statement also provided the Defense Department’s first official confirmation of a partnership with Google, which had been reported earlier in the week but not formally acknowledged.

In its official statement, the Pentagon said the agreements “accelerate the transformation toward establishing the United States military as an AI-first fighting force.”

Why Anthropic Was Excluded

Anthropic’s absence from the deals stems from its designation earlier this year as a supply-chain security risk. The Pentagon has barred the use of Anthropic’s tools across military departments and defense contractors, with current directives mandating the removal of all Anthropic products within a six-month window.

Pentagon Chief Technology Officer Emil Michael addressed the situation on CNBC Friday, confirming that Anthropic “remains a supply-chain risk.” He also commented on Anthropic’s recently released Mythos AI system — which has drawn attention for its advanced cybersecurity capabilities — calling it a “separate national security moment.” Mythos has been distributed to various companies to help defend IT infrastructure against cyberattacks, though it remains unconfirmed whether the Pentagon has access to the preview version.

Despite the official stance, the political picture is more complex. President Trump said last week that Anthropic was “shaping up” in the eyes of his administration, a comment that has fueled speculation about a possible reversal of the company’s blacklisting.

The Pentagon’s Own Staff Disagrees

Perhaps the most striking aspect of this story is the internal resistance. Military personnel, former defense officials, and IT contractors have told Reuters they are reluctant to abandon Anthropic’s tools. Many view Claude and related Anthropic products as superior to the alternatives now available.

This creates an unusual situation: the Pentagon is officially phasing out a technology that its own workforce considers best-in-class. The department’s stated rationale for bringing in seven vendors is to avoid “vendor lock” — a reference to its heavy prior reliance on Anthropic’s systems.

The Pentagon wants to diversify its AI supply chain. But what happens when the diversified options don’t match the quality of the one you’re replacing?

How the Military Is Actually Using AI Today

The scope of military AI adoption is already substantial. The Pentagon’s GenAI.mil platform has been used by over 1.3 million Defense Department personnel in just five months of operation. Military units are leveraging AI tools across multiple operational domains:

Planning and logistics: AI assists in coordinating supply chains, resource allocation, and deployment scheduling for large-scale operations.
Target identification: Machine learning models process sensor data and satellite imagery to identify and prioritize targets.
Intelligence analysis: Natural language processing systems sift through vast volumes of intercepted communications and open-source intelligence.
Cybersecurity operations: AI-driven tools monitor network traffic for anomalies and respond to potential threats in real time.

The rapid adoption — 1.3 million users in five months — suggests that AI has moved from experimental to essential within the Defense Department.

The Seven Companies and What They Bring

Each of the seven selected vendors occupies a distinct position in the AI ecosystem:

Nvidia: Provides the GPU hardware and software infrastructure that powers most AI training and inference workloads globally. Their role likely centers on providing compute infrastructure for classified AI systems.
Microsoft: Offers Azure AI services and has deep existing contracts with the Defense Department through its Integrated Visual Augmentation System (IVAS) and other programs.
Amazon Web Services: Operates the Joint Warfighting Cloud Capability (JWCC) contract and provides cloud infrastructure for classified workloads.
SpaceX: Brings satellite communications and Starlink connectivity capabilities, which are increasingly important for AI-enabled operations in contested environments.
OpenAI: The creator of ChatGPT and GPT-4, offering general-purpose AI models that can be adapted for military applications.
Google: Provides cloud AI services and has a history of defense contracts, including Project Maven (though that partnership was controversial and ended).
Reflection AI: A lesser-known player in the group, suggesting the Pentagon is also betting on emerging AI companies.

The Broader Implications

This announcement signals several important trends in the intersection of AI and national security:

Multi-vendor strategy over single-provider dependence. The Pentagon’s explicit goal of avoiding “vendor lock” reflects lessons learned from its heavy reliance on Anthropic. By contracting with seven companies, the department aims to maintain flexibility and competitive pricing.

The classified AI market is now real. Impact Level 6 and 7 certifications are not trivial. Companies must meet rigorous security standards to operate in these environments. The fact that seven companies have now achieved this status suggests a maturing defense AI market.

Anthropic’s future in defense remains uncertain. While currently excluded, the company’s technology is reportedly valued by military personnel, and political signals from the White House suggest the situation could change. Anthropic is also reportedly weighing a new funding round at a valuation near $900 billion, according to recent reports.

AI is becoming a battlefield capability, not just a back-office tool. The Pentagon’s vision of an “AI-first fighting force” suggests AI will be integrated into operational decision-making at every level, from strategic planning to tactical execution.

What Comes Next

The next few months will be critical. The Pentagon must manage the six-month transition away from Anthropic’s tools while integrating seven new AI providers into its classified networks. Military personnel will need training on new systems, and the department will need to ensure interoperability between multiple AI platforms.

Meanwhile, the question of whether Anthropic’s exclusion is permanent or temporary remains open. With political signals suggesting a possible reversal and military personnel advocating for the company’s tools, this story is far from over.

For the broader AI industry, the Pentagon’s multi-vendor approach could serve as a model for enterprise AI adoption — diversifying risk while maintaining access to cutting-edge capabilities. But it also raises questions about whether spreading contracts across seven providers dilutes the depth of integration that any single vendor could achieve.

One thing is clear: the race to become the U.S. military’s primary AI partner has just intensified, and the stakes have never been higher.

Pentagon Signs Classified AI Deals with Seven Tech Giants — Leaving Anthropic on the Sidelines