ChatGPT accounted for roughly 90% of all AI spending by US House offices, committees, and institutional accounts during the year ending March 31, according to House disbursement records reviewed by CNBC. Congress spent $100,580 on OpenAI's product across 798 transactions, out of $113,740 in total AI spending — a lopsided split that puts every other vendor in the single digits.
Claude came in second, and it was not close. Anthropic pulled $13,160 across 37 transactions, roughly one-eighth of ChatGPT's take and about 12% of total AI outlays. Every other paid AI tool combined for well under $1,000 of the disclosed spend.
The partisan split is the other headline number. Democratic offices spent $54,165 on AI tools, three times the $15,782 recorded by Republican offices. The disbursement records do not explain the gap, but the raw ratio is a rare quantitative snapshot of adoption behavior on the Hill.
Key facts
- 01OpenAI captured roughly 90% of House AI spending, per disbursement records for the year ending March 31.
- 02Congress spent $100,580 on ChatGPT across 798 transactions, out of $113,740 in total AI purchases.
- 03Anthropic's Claude placed second with $13,160 across 37 transactions — about an eighth of ChatGPT's take.
- 04Democratic offices spent $54,165 on AI tools, three times the $15,782 spent by Republican offices.
- 05Figures exclude free accounts and AI features bundled into broader software contracts.
The figures come with a meaningful caveat: they capture only line-item purchases in House disbursement filings. Free ChatGPT and Claude accounts are invisible in the data, as are AI features bundled into broader software contracts — Microsoft 365 Copilot licensed through an enterprise agreement, for instance, or Gemini rolled into a Google Workspace subscription. Actual AI usage across congressional staff is almost certainly higher than $113,740 suggests.
What staff are doing with the tools is the mundane middle-office work of a legislative office. Congressional staffers are using ChatGPT and other tools to write memos, summarize and analyze legislation, respond to constituents, prepare hearing materials, sort through policy research, and even draft social media posts. None of those tasks is unique to government — they mirror the paralegal-and-analyst use cases that have driven ChatGPT adoption inside law firms, consultancies, and corporate policy teams.
For OpenAI, a 90% share of a single institutional buyer matters less as revenue than as distribution. Congress is a reference customer whose staff cycle in and out of federal agencies, campaigns, K Street firms, and the private sector. Every memo drafted with ChatGPT in a member's office is a habit that follows the staffer to their next job.
For Anthropic, the second-place finish is consistent with its broader market position: a strong developer and enterprise story, a weaker foothold in the general-purpose knowledge-worker seat where ChatGPT's brand dominance is hardest to dislodge. Claude Sonnet and Opus are heavily used inside coding workflows and via API integrations that would not show up as a Claude.ai subscription line item, which likely undercounts Anthropic's real presence on the Hill.
The data also lands in the middle of an active policy debate about AI in government. The White House and multiple agencies have issued guidance on federal AI use in the past year, and members of Congress have introduced bills covering everything from AI-generated deepfakes to procurement standards. Legislators writing rules for the technology are, per their own expense reports, also paying customers of the largest company they are writing rules for.
There are limits to what the numbers show. $113,740 is a rounding error against the House's operating budget, and paid seat counts do not equate to policy influence or output quality. The records also do not capture the Senate, executive-branch agencies, or state legislatures, all of which run their own procurement.
Still, when a single vendor takes 90% of a category inside one of the most-watched buyers in Washington, that is a distribution moat worth tracking. OpenAI's consumer brand has translated into institutional default status faster than most enterprise-software categories manage in a decade. The question for Anthropic, Google, and Microsoft is whether bundled-in Copilot and Gemini seats — the spending the House records cannot see — are quietly closing the gap, or whether ChatGPT's head start on the Hill compounds the same way it has everywhere else.
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.
