Overview
Government records and interviews with federal employees and contracting experts indicate that xAI’s Grok chatbot has seen only minimal usage within civilian federal agencies. The Office of Management and Budget, which collected a consolidated 2025 inventory of agency AI deployments, lists more than 400 publicly identified instances where a specific vendor was named. Of those entries, only three involve xAI or Grok.
That low count stands in stark contrast to other vendors. The OMB data show 234 instances naming technology based on OpenAI models such as ChatGPT, Codex, and Microsoft Copilot. Alphabet products, including Gemini, are cited in 33 entries, and Anthropic’s Claude appears in 26, although Claude has since been blacklisted by the prior administration.
Officials with the Office of Management and Budget did not respond to requests for comment on the inventory. xAI did not provide answers to detailed questions about Grok’s federal usage. Many other companies whose products appear in the inventory also did not reply to requests for comment. Google declined to provide a comment and pointed to public posts describing its government engagements.
Pricing, procurement and adoption dynamics
Grok has been available to federal agencies through the General Services Administration for about eight months at an introductory cost of 42 cents per agency. That near-zero pricing mirrors strategies used by several AI vendors, which offer low-cost or free access initially to encourage agency experimentation and create pathways toward larger and higher-priced enterprise contracts.
"The goal is to encourage adoption so that federal employees eventually can’t imagine doing their jobs without generative AI," said Valerie Wirtschafter, a researcher at the Brookings Institution who studies AI adoption in the federal government.
OMB’s consolidated inventory offers a snapshot that raises questions about Grok’s ability to wrest meaningful market share from established competitors and thereby support the revenue projections used in SpaceX’s own financial statements. In a regulatory filing, SpaceX outlined expectations that its AI work could be the largest revenue driver among its businesses, valuing the total opportunity for AI services to large companies and organizations at $26.5 trillion. The company’s proposed initial public offering plans contemplate a valuation of $1.75 trillion.
Analysts and executives outside SpaceX warn that weak public-sector uptake may be an early signal of broader commercial challenges. "It suggests the model lacks the security rigor required at the federal level, which will be a red flag" for some corporate buyers, said Vineet Jain, co-founder and CEO of Egnyte. He added, "Without government validation, the $1.75 trillion valuation looks less like a floor and more like a high ceiling."
Leadership advocacy and internal promotion
SpaceX’s chief executive has been vocal about Grok’s potential for federal use and has advocated for rapid deployment across government. During the announcement of Grok’s availability through the GSA, he expressed a desire to work with the administration to expand AI deployment throughout the federal government.
At one point an internal SpaceX office called the Department of Government Efficiency, or DOGE, actively promoted Grok across agencies. Reports show that DOGE encouraged use of Grok within the Department of Homeland Security, even in cases where the chatbot had not been formally approved for use by that agency.
What the inventory reveals about use cases
The OMB inventory provides descriptions of how agencies use AI tools and how many employees are involved in each use case. Entries range from routine tasks, such as sorting emails and transcribing meetings, to more advanced applications like fraud detection or space research. National security-related use cases are typically omitted from the public inventory.
The data contain inconsistencies. Some agencies left the specific service field blank on the form, and definitions of what qualified as an AI use case varied across agencies. Wirtschafter cautioned about those variances, while noting the dataset is still the most comprehensive non-military, non-intelligence inventory available for federal AI use.
Where Grok does appear in the inventory, its deployments are modest and largely confined to low-complexity tasks. For example, the Office of Personnel Management and the Department of Health and Human Services listed Grok for elementary functions such as drafting initial versions of documents or preparing social media posts. HHS did not respond to queries about its usage. An OPM spokesperson said Microsoft Copilot is the agency’s most commonly used AI tool.
A separate portion of the inventory that catalogs more ambitious, higher-impact applications likewise shows little trace of Grok. The three entries tied to Grok describe limited tests or pilot deployments at the Department of Energy’s Lawrence Livermore National Laboratory and at the Election Assistance Commission. By comparison, OpenAI and Microsoft together account for 140 use cases in that segment of the inventory. The Energy Department did not reply to requests for comment. The Election Assistance Commission said its evaluation of the pilot was ongoing.
Defense and military exceptions
The OMB inventory does not cover defense-related systems. Separate engagements with the Department of Defense show a different picture. The Pentagon has a contract valued at approximately $200 million with xAI. Earlier in the year the Defense Secretary announced Grok’s addition to GenAI.mil, which is the military’s unclassified hub for AI model use. xAI is also among a group of firms that have been authorized to deploy models on the Defense Department’s classified networks.
Despite these formal avenues, some defense personnel report preferring other models to Grok. One Pentagon source said many staffers favored competitor tools for their work. At the Defense Advanced Research Projects Agency, or DARPA, the source reported that Google’s Gemini is used for engineering analysis, while Anthropic’s Claude is favored for coding, writing and research. OpenAI tools are also used at DARPA, but Grok is generally not, according to that source, who described Grok as "just not the best model out there." The Pentagon and DARPA did not comment for this report.
Ongoing efforts and recent procurement outcomes
xAI has taken steps to win broader federal acceptance. The company has sought FedRAMP High Authorization, the security designation required for handling sensitive government data, and is pursuing that approval with sponsorship from the U.S. Department of Agriculture. Yet three IT professionals at USDA said they were not aware of Grok being used within their agency. USDA stated it was "proud to sponsor Grok" but did not specify how widely the tool is used.
Other procurement outcomes suggest Grok has not always met agency requirements. A person familiar with a recent competition said xAI lost a bid to develop a Grok-powered product for the Department of Veterans Affairs because the chatbot did not meet that department’s needs. The VA did not directly answer questions about any Grok evaluations.
Corporate adoption and usage patterns
Federal usage trends are mirrored by evidence in the private sector showing limited enterprise adoption of Grok. Netskope, a web traffic monitoring firm that tracks how corporate customers connect to AI models, published a report last year noting that Grok had "failed to gain significant traction" in corporate environments. Updated metrics provided to Reuters by Netskope show Grok enterprise usage declined from a peak of five users per 1,000 to two users per 1,000.
Netskope’s executive Ray Canzanese said employees who did use Grok engaged with the chatbot for shorter periods relative to competitors. He noted Grok users spent less than half the time that ChatGPT users spent with OpenAI’s model. Canzanese concluded that the available usage data suggested Grok "is just not going to enter the mainstream for corporate America."
Implications and open questions
The limited footprint of Grok across civilian federal agencies, combined with weak corporate uptake indicators, creates uncertainty about xAI’s ability to scale into the kind of enterprise business that SpaceX has identified as central to its AI revenue forecasts. SpaceX’s public filings present AI as a major growth vector, and the company projects a vast total addressable market. The federal inventory data and independent corporate usage metrics offer a reality check on those projections, while also reflecting variability in agency definitions and reporting.
At present, Grok has achieved targeted defense contracts and some authorized deployments on classified networks, but adoption across civilian government and in corporate America remains modest. The available evidence highlights gaps in both security vetting and user engagement that xAI will need to address if Grok is to become a broadly used enterprise AI product.