A 1-hour trading approach

On August 31, the Department of Defense switched on two new tools inside GenAI.mil, its Common Access Card-gated enterprise AI platform: OpenAI’s ChatGPT Mil and Starshield AI’s Grok for Government, joining the Google Gemini tool that started the platform when it launched on December 9, 2025. The rollout hands frontier AI to a workforce of more than 3 million.

This is not a procurement footnote. It is a snapshot of which companies are compounding revenue and relationships inside the most durable customer the technology industry has ever known, and which one is still locked out.

What Was Switched On

ChatGPT Mil is available for work involving controlled unclassified information. The Department of Defense has said the tool is accredited at Impact Level 5, a DoD cloud-security standard for certain sensitive unclassified information. The department and OpenAI have also described the deployment as running inside government cloud infrastructure and positioned it as a government environment rather than the public ChatGPT service.

OpenAI is providing ChatGPT Mil through its OpenAI for Government business, while Grok is being delivered through Starshield AI, a SpaceX unit. Both companies received contracts with ceilings of up to $200 million each from the Pentagon’s Chief Digital and Artificial Intelligence Office in July 2025. Google and Anthropic received the same ceiling under those initial awards.

GenAI.mil was first launched with Google’s Gemini for Government. By early July 2026, Defense Department officials said the platform had reached about 1.7 million users and that users had created more than 100,000 custom agents. Monday’s additions extend that footprint further.

The Absent Company

Notably absent from Monday’s addition of Grok and ChatGPT to GenAI.mil is Anthropic’s Claude.

The company’s exclusion carries real context. Anthropic sued the Pentagon in federal court in California after the Defense Department designated Anthropic a national security supply chain risk. News reports have linked the dispute to Anthropic’s refusal to allow unrestricted military use of its technology, including objections tied to domestic surveillance and autonomous weapons.

The court sided with Anthropic. U.S. District Judge Rita Lin ruled that the Pentagon’s designation was unlawful, calling it “illegal and baseless” and writing that the invocation of national security was not a blank check to retaliate against critics.

But court victories do not automatically translate into contract slots. The Pentagon has not ruled out an appeal. Meanwhile, OpenAI and xAI are embedding deeper by the week.

Who Else Benefits

Palantir sits in a different position entirely. Project Maven has increasingly shifted from experimentation toward operational use, and Palantir continues to position Maven Smart System as a scaled program inside the Department of Defense.

Microsoft’s role is structural rather than direct: OpenAI runs on Azure government cloud, so every ChatGPT Mil session travels across Microsoft infrastructure. The revenue is less visible than a named contract, but it is durable.

The Wider Point

The Pentagon’s rollout highlights a broader shift in how major AI companies are approaching government customers. Instead of treating public-sector agencies as another enterprise account, companies are increasingly developing specialized versions of their models designed around government requirements, security controls, and deployment environments. That specialization builds switching costs on both sides.

Anthropic won in court. OpenAI and xAI won in the field. For investors tracking who controls the most consequential enterprise AI deployment of this decade, the distinction matters.