--- title: "HHS banned Claude in March. Its October fix is a no-bid Anthropic award after a $1-a-year run" description: "A notice filed October 19 would keep the Department of Health and Human Services on Anthropic's Claude without competition, past the October 31 expiry of a $1-a-year government deal and seven months after President Trump ordered the model phased out of agencies. The price is undisclosed; the company may file IPO papers within days." publisher: "The Inference" section: "Policy & Labor" published: 2026-10-23T13:18:25.869Z modified: 2026-10-23T13:18:25.869Z canonical: https://theinference.org/article/hhs-banned-claude-in-march-its-october-fix-is-a-no-bid-anthropic-award-after-a-1-a-year-run language: en keywords: "Anthropic, AI Policy, Government Contracts, Lobbying, IPO" --- # HHS banned Claude in March. Its October fix is a no-bid Anthropic award after a $1-a-year run > A notice filed October 19 would keep the Department of Health and Human Services on Anthropic's Claude without competition, past the October 31 expiry of a $1-a-year government deal and seven months after President Trump ordered the model phased out of agencies. The price is undisclosed; the company may file IPO papers within days. ## A no-bid award, seven months after a ban On October 19 the Department of Health and Human Services posted [a notice on SAM.gov](https://www.beckershospitalreview.com/healthcare-information-technology/ai/hhs-plans-no-bid-enterprise-ai-deal-with-anthropic/), the federal contracting site: a sole-source award to [Anthropic](https://theinference.org/markets/companies/anthropic), maker of the Claude AI models, to keep the department's workforce on the tools without open competition, with no contract value or award date listed [1]. The justification is continuity itself: "Claude is Anthropic's proprietary model family, and the requirement is continuity of the existing Claude deployment already," the notice says [2]. Under its terms, HHS data is excluded from model training by default, stays in the continental United States, and Anthropic must report security incidents within one hour [1]. ## From $1 a year to a phase-out, and back HHS rolled Claude out department-wide in December 2025, after Anthropic offered agencies the service for $1 a year under the General Services Administration's OneGov program [1][2]. March brought the reversal: talks over military use collapsed, the Pentagon designated Anthropic a supply chain risk, President Donald Trump ordered agencies to phase the company out, and HHS told employees to stop using Claude while the FDA moved its Elsa chatbot to Google's Gemini [1][2][3]. A federal appeals court [upheld the Pentagon label 2-1 on September 25](https://theinference.org/article/anthropic-s-pentagon-ban-stands-on-appeal-and-its-ipo-carries-518-billion-of-locked-compute-bills), and Anthropic asked the full court on October 13 to reconsider [3][4]. The new notice never mentions the ban [1]. Anthropic, for its part, made Claude for Government generally available this month and hired a federal civilian sales chief on October 21 [5]. ## The dependency admission, and its lobbying bill Jessica Tillipman, associate dean for government procurement law studies at George Washington University, told [FedScoop](https://fedscoop.com/hhs-sole-source-award-anthropic-ai-claude/) the notice "appears to be an admission that the agency is dependent on Claude," and that general chatbot services are a questionable basis for skipping competition: "this is going to be a pain, so let's just keep it going" [2]. The $1 offer did what free samples are built to do [6]. The same quarter set Anthropic's [lobbying record at $2.1 million](https://www.theverge.com/policy/1016585/openai-anthropic-nvidia-lobbying-records), roughly double its previous best, and still behind [Nvidia](https://theinference.org/markets/companies/nvidia)'s $2.2 million and [Apple](https://theinference.org/markets/companies/apple)'s $3.1 million [6]. *Chart: **Anthropic doubled its lobbying record in Q3 and still trailed Nvidia and Apple** Bar chart of Q3 2026 federal lobbying spending: Apple $3.1M, Nvidia $2.2M, Anthropic $2.1M, OpenAI $1.5M* *Federal lobbying disclosures for July to September 2026, in millions of US dollars, as compiled by The Verge. [6]* ## Pay real money or migrate, days before the prospectus OneGov expires October 31, so HHS now pays a real, undisclosed price or migrates; [OpenAI](https://theinference.org/markets/companies/openai) has already swapped its own $1 deal for consumption pricing through 2028 [7]. The money runs from taxpayers to a private company that could [file IPO papers as early as Friday](https://www.msn.com/en-in/news/other/anthropic-ipo-filing-5-things-to-know/ar-AA2fQpbO), toward [a listing discussed at a $2 trillion valuation](https://theinference.org/article/anthropic-pitches-a-2-trillion-ipo-over-the-pentagon-s-standing-removal-order), on first-half 2026 revenue of $16.3 billion and a net loss above $80 billion [8]. [The Inference reported in September](https://theinference.org/article/anthropic-is-betting-the-enterprise-wont-blink-at-the-invoice) that Anthropic was betting enterprises would not blink at the invoice; HHS has now filed that bet in its own name. Three readings settle it: the award value when it posts, whether the competitive frontier-model contract actually opens, and the court's answer on reconsideration. ## Takeaway Seven months after the White House ordered Claude phased out of agencies, HHS is skipping competition to keep it, filing what a procurement law scholar calls an admission of dependency. The undisclosed award price, a promised broader competition and the court's answer on the Pentagon label now decide what the reversal costs. ## Sources 1. [Becker's Hospital Review, "HHS plans no-bid enterprise AI deal with Anthropic," 22 October 2026](https://www.beckershospitalreview.com/healthcare-information-technology/ai/hhs-plans-no-bid-enterprise-ai-deal-with-anthropic/) 2. [FedScoop, "HHS to make sole-source award to Anthropic ahead of broader AI competition," 20 October 2026](https://fedscoop.com/hhs-sole-source-award-anthropic-ai-claude/) 3. [CNBC, "U.S. appeals court upholds Pentagon designation of Anthropic as supply chain risk," 25 September 2026](https://www.cnbc.com/2026/09/25/pentagon-anthropic-ai-risk-appeals-court.html) 4. [Crypto Briefing, "Anthropic asks full appeals court to revisit Pentagon national security ruling," 13 October 2026](https://cryptobriefing.com/anthropic-appeals-court-national-security-ruling/) 5. [GovCon Wire, "Anthropic Names Brian Mikkelsen Federal Civilian Go-to-Market Head," 21 October 2026](https://www.govconwire.com/articles/anthropic-brian-mikkelsen-federal-civilian-gtm-head-hired) 6. [The Verge, "OpenAI, Anthropic, and Nvidia had record lobbying quarters," 23 October 2026](https://www.theverge.com/policy/1016585/openai-anthropic-nvidia-lobbying-records) 7. [FedScoop, "Anthropic extends its GSA OneGov deal for Claude by one month," 17 September 2026](https://fedscoop.com/anthropic-extends-gsa-onegov-deal-for-claude-by-one-month/) 8. [The Economic Times via MSN, "Anthropic IPO filing: 5 things to know," 22 October 2026](https://www.msn.com/en-in/news/other/anthropic-ipo-filing-5-things-to-know/ar-AA2fQpbO) --- HHS banned Claude in March. Its October fix is a no-bid Anthropic award after a $1-a-year run — The Inference. Canonical: https://theinference.org/article/hhs-banned-claude-in-march-its-october-fix-is-a-no-bid-anthropic-award-after-a-1-a-year-run