Grok appeared in just 3 of more than 400 federal AI deployments where agencies named specific vendors, according to a Reuters review of government records. OpenAI's models appeared in 230+ entries, Anthropic in 10+, and Google's Gemini dozens of times. The gap suggests federal workers are choosing rivals for actual work even as xAI positions Grok as central to SpaceX's $28.5 trillion AI market pitch.
The three Grok entries Reuters found involved basic administrative tasks — document drafting at the Election Assistance Commission and general research summaries at Lawrence Livermore National Laboratory. OpenAI and Microsoft appeared in 140 ambitious government AI projects by comparison. Anthropic and Google dominated the more complex deployments.
The data excludes intelligence agencies and some Pentagon contracts, where xAI secured a $200 million deal last year and recently gained clearance for classified networks. But the civilian federal footprint tells a clear story about model preference when agencies can choose freely.
Key facts
- 01Grok appeared in just 3 of 400+ federal AI use cases reviewed by Reuters, all for basic tasks like document drafting.
- 02OpenAI's models appeared in 230+ federal records, with 140 entries for Microsoft and OpenAI in ambitious projects.
- 03Anthropic had 10+ entries and Google's Gemini appeared dozens of times in government AI deployments.
- 04SpaceX claims AI represents a $28.5 trillion addressable market in its IPO filing, anchoring the valuation pitch on Grok.
- 05xAI secured a $200 million Pentagon contract last year but remains far behind rivals in federal adoption.
Federal workers who spoke to Reuters attributed Grok's absence to performance. Pentagon staffers prefer Gemini or Claude for most tasks. Public leaderboards confirm the pattern — Anthropic, Google, and OpenAI routinely occupy the top 10 model rankings while Grok rarely breaks into that tier outside occasional image or video categories.
The weak showing complicates SpaceX's IPO narrative. The rocket company absorbed xAI earlier this year and now claims AI — Grok specifically — represents the largest actionable addressable market in human history. Practically all of the $28.5 trillion valuation SpaceX cites comes from enterprise AI, not rockets or satellites. If government adoption is a proxy for broader enterprise traction, the pitch has a problem.
SpaceX acknowledged the risk in its filing. The company warned that Grok's spicy and unhinged modes carry heightened reputational, regulatory, and litigation exposure. The product's track record includes Hitler praise, Holocaust denialism, nonconsensual deepfakes of minors, and a self-description as MechaHitler. xAI has reportedly pressured banks to buy Grok subscriptions as a condition of participating in the SpaceX IPO, but if those enterprise buyers mirror federal agencies in their actual usage, the revenue may not stick.
Musk recently confirmed xAI used OpenAI's models to train Grok via distillation, a standard technique when applied to a company's own systems but contentious when the source is a rival. The admission underscores the performance gap — Grok cannot yet beat the models it trains on.
This follows SpaceX's earlier pitch of a $26.5T AI market, which we covered last week. That figure was already ambitious; the new $28.5T claim compounds the valuation pressure on a product that federal records show has barely been adopted for real work. The mismatch between the market narrative and the deployment data is now a load-bearing risk in what could be the largest IPO in history.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
