OpenAI Swoops as Anthropic Stumbles with Pentagon

In the high-stakes world of AI and defense contracts, OpenAI has emerged as the victor after Anthropic's negotiations with the Pentagon unraveled. Internal messages reveal that while OpenAI CEO Sam Altman portrayed himself as a peacemaker, securing the Pentagon contract suggests otherwise. This isn't just a partnership announcement. It's a convergence.

Behind the Curtains

The drama unfolded over a tense 10-day period, culminating on March 2. On February 24, Altman addressed worried employees in a Slack channel, suggesting it was an opportune moment to negotiate a deal for classified use of OpenAI's systems with Defense Under Secretary Emil Michael. By the next day, draft contract language was already being exchanged.

The AI-AI Venn diagram is getting thicker. Altman's comments on February 26 highlighted his conflicting roles. He expressed a desire to de-escalate tensions while still angling to secure a deal. Although he admitted the optics might be unfavorable, he emphasized principle over appearance.

A Rivalry Intensifies

By February 27, Altman disclosed to key staff that negotiations between the Pentagon and Anthropic had soured, partially due to Anthropic CEO Dario Amodei's media strategies. As a deadline approached, Altman was informed that the Pentagon could offer Anthropic a way out of its supply chain risk label. Still, Altman was working hard to 'save' a competitor who he felt had been undermining him for years.

When the dust settled, OpenAI had secured the contract, but not without public relations consequences. Altman framed his actions as altruistic, but many on the Anthropic side, including Amodei, saw through this. Amodei even labeled Altman's peacemaker narrative as dishonest in a leaked memo, though he later apologized. If agents have wallets, who holds the keys?

The Path Ahead

Anthropic remains in the mix for Pentagon work, but a March 2 message from Altman highlighted a significant oversight. OpenAI's contract included a clause requiring separate negotiations for deploying ChatGPT in intelligence agencies. The Pentagon confirmed OpenAI's deal would be offered to others, but not Anthropic, due to its existing agency integrations.

This raises a critical question: Is Altman's strategy more about altruism or simply outmaneuvering a rival? The compute layer needs a payment rail, and OpenAI seems intent on laying it down. The eventual impact on Anthropic could reshape its standing in the AI sector, but for now, OpenAI has seized a moment that could redefine its trajectory.