corporate VERIFIED
Feb 23, 2026, 00:00 UTC Pentagon

xAI Signs Classified Military AI Deal

Elon Musk's xAI signs a deal to deploy Grok in classified military systems as part of the Pentagon's $200M Other Transaction Authority program, one day before Hegseth's ultimatum to Anthropic.

On February 23, 2026, Elon Musk’s xAI signed a contract to deploy its Grok model on classified military networks as part of the Pentagon’s $200 million Other Transaction Authority (OTA) AI program — the same program from which Anthropic would be expelled four days later. The deal was announced one day before Defense Secretary Hegseth issued his ultimatum to Anthropic, placing the xAI agreement in a timeline that suggested the Pentagon was lining up replacements before the confrontation with Anthropic reached its climax.

The xAI deal covered deployment of Grok for intelligence analysis, logistics optimization, and operational planning on networks classified up to TOP SECRET/SCI. Unlike Anthropic’s Claude, Grok had no published safety framework restricting military applications, and xAI had made no public commitments to red lines on autonomous weapons or mass surveillance. The contract’s terms, insofar as they were publicly disclosed, included no provisions for use-case restrictions or AI safety guardrails — precisely the arrangement the Pentagon had demanded from Anthropic and been refused.

The timing raised conflict-of-interest questions. Musk’s companies — SpaceX, Starlink, and now xAI — represented a growing share of the Pentagon’s technology supply chain. Critics argued that the Trump administration’s willingness to punish Anthropic while rewarding xAI reflected Musk’s political relationship with the White House rather than a neutral assessment of technology capabilities. Grok’s performance on classified military benchmarks was not publicly known, and independent evaluations of the model consistently ranked it below Claude on reasoning and analytical tasks.

Defense procurement analysts noted that the xAI deal, combined with the OpenAI agreement that would follow four days later, represented a deliberate strategy to fragment the classified AI market across multiple providers. This would reduce any single company’s leverage in future negotiations — ensuring that no AI company could ever again do what Anthropic had done: refuse Pentagon demands from a position of sole-provider strength.

Sources