Primary Thought
Government & Public Sector AI

HHS Plans Sole-Source Bridge Deal for Anthropic's Claude

HHS says a six-month sole-source bridge will keep Claude running while it prepares a competition. Procurement lawyers question the justification.

Priya Raman
Priya Raman
November 5, 2026
Share
Anthropic Claude AI illustration showing the US Department of Health and Human Services (HHS) proposed sole-source contract, government AI procurement and enterprise AI services.

The US Department of Health and Human Services plans to award Anthropic a sole-source contract to keep Claude running inside the agency while a competitive procurement is prepared. The notice, reported by FedScoop on October 20, 2026, has drawn skepticism from procurement lawyers who question whether only one vendor can meet HHS's needs. It is a useful test case for how federal agencies handle vendor dependence when promotional AI pricing ends.

What HHS has said

According to FedScoop's first report, HHS posted a notice on SAM.gov describing a "limited continuity of services award" to Anthropic for Claude. The agency says the award is an interim measure while it prepares a competitive purchase. The notice lists the functions involved as chat, drafting, summarization, data-analysis workflows, application access, and administrative controls, and says the requirement is continuity of the existing Claude deployment. No dollar value appears in the notice.

A contracting-news summary gives the notice number as 7571TE27R00001 and quotes the framing that the capability is specific to one provider. It also reports that no award date or period of performance is stated in the notice itself. A follow-up FedScoop article from October 23 adds that an HHS spokesperson described the contract as a temporary six-month sole-source bridge, intended to cover the gap when Anthropic's discounted federal offering expired on October 31, and that HHS aims to have longer-term arrangements executed before the bridge ends.

HHS also intends to buy enterprise AI services through a competitive multiple-award blanket purchase agreement. The contracting-news summary says no solicitation date or award date has been published, so contractors should treat that vehicle as planned rather than open.

Background: from a $1 offer to a bridge

FedScoop reports that HHS rolled out Claude to its workforce in December 2025, after Anthropic offered its services to agencies for $1 a year through the General Services Administration's OneGov program. That offer was extended by a month and was due to end October 31. According to FedScoop, earlier in 2026 a clash between Anthropic and the Pentagon led to a governmentwide ban on Anthropic's tools; HHS disabled its Anthropic products, and the FDA began moving its Elsa chatbot to Google's Gemini. The October 23 article says a federal judge has since enjoined the ban, and that the legal dispute is ongoing.

The pricing picture has also moved. The Next Web reported that on October 27, Anthropic announced a new GSA agreement giving 40% off token-based usage across its Claude models for federal agencies and, for the first time, state, local, tribal and territorial governments, with no minimum spend. Anthropic estimated the discounts could save the government $1 billion. The same article says OpenAI has a comparable GSA deal at 50% off token usage. The Next Web, citing a GSA website, puts the new Anthropic agreement's term at 27 months, to January 31, 2029; the exact term should be checked against GSA.

Why procurement experts are skeptical

Sole-source awards are permitted when only one vendor can meet a requirement, and that is the claim being tested. Jessica Tillipman of George Washington University told FedScoop that the notice reads like an admission of dependence on Claude. She said the described services sound like general chatbot functionality, which makes the claim that only Claude can provide them questionable, and that the rationale amounts to convenience: switching is a hassle, so the agency stays. In the follow-up article, she called the reasoning circular, since the notice effectively says that only Anthropic can provide Claude, and she asked why HHS could not prepare for a transition whose end date was public.

David Timm of the law firm Burr and Forman told FedScoop that the uniqueness and continuity arguments are "tough sells," noting that HHS stopped using Anthropic during the ban and apparently managed with other models. He said those points could give a bid protest footing, though he could not judge the odds without a fuller record. Competitors such as OpenAI or Google could in theory protest, but because they already sell to HHS, it is unclear whether they would.

The value question is also unresolved. FedScoop reports that the award appears to rely on simplified acquisition threshold authority, which would imply a value below $350,000, but Timm said it is unclear that authority applies since the notice gives no cost. Larger sole-source awards require a formal justification and approval document, which the notice lacks. HHS did not provide the sole-source justification FedScoop requested.

The lock-in lesson

The episode illustrates a risk Tillipman has written about: when AI tools become embedded in daily workflows, switching costs can turn a promotional price into long-term dependence. Natalie Buda Smith, chief AI officer at the Library of Congress, urged agencies in FedScoop's reporting to stay flexible rather than commit to one approach, arguing that AI tools behave more like utilities than capital purchases.

For agencies, the practical lessons are familiar from other technology cycles but sharper here. A one-year promotional price creates a cliff, and a deployment that grows during that year can make a later competition look impractical. Contract terms that preserve data portability, prompt and workflow exportability, and the option to run more than one model may reduce the need for exceptions of this kind.

What to watch next

  • The justification document. Whether HHS publishes or defends its sole-source rationale, and whether it uses the cost threshold authority commentators have speculated about.

  • A protest. Any challenge from a competitor or interested party would test the continuity argument formally.

  • The competitive agreement. The timing and structure of the multiple-award purchase agreement will show whether other providers get a genuine opening.

  • The ban litigation. The outcome of the dispute over the governmentwide restriction affects every agency using Claude.

What is confirmed is the existence of the notice, its stated scope, and the six-month bridge description from an HHS spokesperson. What remains contested is whether the sole-source justification holds. Nothing in the sources reviewed shows whether the award has been finalized, so readers should treat it as planned until HHS or Anthropic says otherwise.

Frequently Asked Questions

What is HHS planning?

A limited sole-source "continuity of services" award to Anthropic for Claude, described by an HHS spokesperson as a temporary six-month bridge, per FedScoop.

Why does it matter?

Critics say the justification that only one vendor can provide chatbot functions is weak, and the case highlights vendor lock-in after promotional pricing ends.

What comes next?

HHS plans a competitive multiple-award blanket purchase agreement for enterprise AI, but no solicitation date has been published.

What are the new GSA prices?

The Next Web reports Anthropic announced 40% off token-based usage for governments on October 27, versus a reported 50% in OpenAI's GSA deal.

Related articles