OpenAI, Google, and Perplexity are close to securing approval to sell their AI technologies directly to the U.S. government via their own cloud platforms, according to an insider. This approval, expected to be on a “low impact” pilot basis, represents a significant move toward operational independence. Historically, AI firms have depended on established government contractors like Microsoft, Palantir, and AWS to host their services for federal use, which limited their control over deployment. In contrast, Anthropic’s reliance on partners like Palantir has sparked complications, exemplified by a Pentagon dispute over AI use in military contexts and concerns about autonomous weapons. Seeking autonomy, OpenAI, Perplexity, and Google pursued accelerated FedRAMP 20x security reviews last year and appear near approval, positioning themselves to engage the government directly. Anthropic has not taken the same path, though it acknowledges the benefits of direct government service offering.
Back
