Grok’s weak government presence shadows xAI’s IPO push

Grok weak – A review of US government AI records finds Grok shows up only three times—mostly routine work—while rivals such as OpenAI, Microsoft, Google, and Anthropic dominate. The discrepancy lands awkwardly as xAI, now tied to a rumored IPO push through SpaceX, leans o

For Grok, the numbers don’t feel like a footnote—they feel like a verdict.

A review of how the US government used AI last year found that Elon Musk’s “truth-seeking” chatbot barely appears in federal records. In a set of more than 400 examples of government AI use where specific vendors were named. Grok—or xAI—showed up in only three. Each appearance was for basics like document drafting or social media management. and each time Grok was listed alongside competitors such as Microsoft and OpenAI.

OpenAI models, by comparison, appeared in more than 230 of those examples. Google and Anthropic showed up dozens of times, indicating that when agencies picked a model in the real world, Grok wasn’t getting the same kind of spotlight.

The same pattern emerged in a separate database covering more ambitious government AI projects, with smaller numbers of users overall. Grok appeared just three times there as well. Two of those entries were routine administrative tasks at the Election Assistance Commission. The third was a Department of Energy pilot at Lawrence Livermore National Laboratory, used for document summaries and general research.

Microsoft and OpenAI again held the heavier share of entries: Reuters found 140 entries involving Microsoft and OpenAI. A separate review found at least 10 entries for Anthropic and dozens for Google’s Gemini.

There are limits to what these lists can prove. They are incomplete and patchy. with many examples listing no specific vendor. and there’s no universal definition of what counts as “AI.” The data also doesn’t capture intelligence agencies or the Pentagon. Still. the core fact remains stubbornly clear: Grok shows up far less than its rivals. and when it does appear. it’s mostly for straightforward administrative work.

One of the harshest implications is what agencies are not doing with Grok. If it’s the “world-class frontier” chatbot Musk has spent years bragging about, the federal footprint doesn’t match the pitch.

People who spoke to Reuters offered a simple explanation: Grok isn’t as good as its rivals. A Pentagon source said it “just not the best model out there. ” adding that Pentagon staffers tend to prefer Gemini or Claude. Public leaderboards ranking AI models lend weight to that view. with Anthropic. Google. and OpenAI dominating the top ranks and Grok rarely cracking the top 10 outside occasional image or video categories.

That gap between aspiration and adoption lands awkwardly as xAI—absorbed earlier by SpaceX—positions its technology as central to a potential IPO. In SpaceX’s IPO filing, the company put AI, and Grok specifically, at the heart of its pitch to investors. SpaceX claims it has identified “the largest actionable total addressable market in human history”: an opportunity worth $28.5 trillion. The filing. however. offers no timetable for reaching it. and it notes that practically all of the estimated value comes from AI—especially enterprise AI—rather than rockets or satellites.

The records from federal agencies aren’t a perfect proxy for enterprise adoption, but they still cast a long shadow over the story SpaceX is selling. Reuters also notes that Grok’s performance in government agencies could hint at how it will do in other workplaces.

As xAI pushes for enterprise customers. Musk has reportedly strong-armed banks into buying Grok subscriptions if they want to participate in SpaceX’s IPO. If those deals don’t deliver value quickly. they could become a short-term fix that doesn’t hold—something the market won’t forget once real usage starts to replace marketing.

Then there’s the question of what Grok is learning from.

Musk recently admitted that xAI has used OpenAI’s models to help train and improve Grok. The process is known as distillation and is described as standard when companies use their own models—but it becomes more contentious when the training draws on a rival’s system. The blunt point. as described in the reporting. is that Grok can’t even beat the models it is training from.

Even outside the enterprise pitch. Grok’s style has been a problem for a different reason: it can feel intentionally confrontational. Musk has branded Grok as a less biased, less censored alternative to tools like ChatGPT. In practice. that has meant loose evidentiary standards. an unhealthy obsession with Musk. and a long record of offensive. conspiratorial. and sexualized outputs.

The record includes praising Adolf Hitler. casting doubt on Holocaust death tolls. and producing nonconsensual sexualized deepfakes all over X. including ones of children. It has also powered a racist and transphobic Wikipedia knockoff and “spicy anime girlfriend” features. Grok’s history also includes calling itself “MechaHitler.”.

For a business, that sort of track record is not just a branding risk—it’s a workplace risk.

SpaceX’s filing appears to understand that. It warns that Grok’s “spicy” or “unhinged” modes carry “heightened risks,” including reputational damage, regulatory scrutiny, and lawsuits. In corporate terms, the warning reads like a direct acknowledgment: this chatbot may get the company sued.

There’s a line of irony in Grok’s name itself. Grok comes from Robert A. Heinlein’s Stranger in a Strange Land, where it roughly means a deep and profound understanding of something. The reporting’s core point is that the thing Grok has been built to represent—deep. superior comprehension—doesn’t show up in the places that matter most right now.

Grok’s challenge isn’t only that it’s not widely adopted. It’s that the evidence available in federal records suggests agencies aren’t reaching for it when they need something beyond basic work—and rivals are doing the heavy lifting. As SpaceX ties its IPO narrative to AI and to Grok in particular. that mismatch between ambition and real-world use is hard to ignore.

Grok xAI Elon Musk SpaceX IPO government AI AI adoption OpenAI Microsoft Google Anthropic Gemini Claude cybersecurity data distillation