Model

Anthropic cuts Claude 40% for US government buyers a day after filing for what could be a $2 trillion IPO

Days after filing a prospectus that could value it at up to $2 trillion, Anthropic gave every US government buyer 40% off Claude, DeepSeek open-sourced a rival coding agent, and the cloud giants that partner with and invest in Anthropic showed strain in their own results. IPO buyers price all three.

Vincent JiangVincent Jiang · 3 min read
Share
Dario Amodei, co-founder and chief executive of Anthropic
1 / 6Slide 1 of 6
Dario Amodei, Anthropic's co-founder and chief executive, days before the roadshow for what could be a $2 trillion IPO.

On October 27, 2026, a day after filing the 335-page prospectus for what could be the largest stock offering ever, Anthropic$350B — Anthropic, private, latest valuation $350B signed something quieter 57. An expanded OneGov agreement with the General Services Administration, the federal government's central purchasing arm, now cuts Claude by 40% for every government buyer in the country 56.

OneGov is the GSA program that negotiates government-wide discounts with AI vendors. The 40% spans Claude Enterprise, Claude for Government and the Claude API for the first time, reaches state, local, tribal and territorial governments, and carries no minimum spend 16.

The 40% haircut landed inside the pricing window

Anthropic estimates the discounts will save the government $1 billion 1. About 500 federal, state and local organizations already use Claude; weekly Claude for Government users are up roughly eightfold since February, and usage across federal civilian agencies has grown about 20-fold in a year 16.

The terms replaced a promotional $1-per-user offer that expired on October 31, 2026, run for 27 months to January 31, 2029, and follow OpenAI's$500B — OpenAI, private, latest valuation $500B parallel OneGov deal, 50% off the tokens AI models bill by 6. None of the prospectus figures reported so far breaks out government sales, so the cost of the 40% in reported dollars cannot be computed from what is public 7.

The prospectus, filed October 26, 2026, showed an $80.7 billion first-half net loss, $72.4 billion of it a one-time accounting charge, on $16.3 billion of revenue, about 14 times a year earlier 7. Reports put the valuation at up to $2 trillion on a raise of $100 billion or more, with the roadshow from the week of November 9 and a listing as soon as the week of November 16 78.

A free rival for Claude Code shipped the same week

On October 31, 2026, DeepSeek, the Chinese AI lab, released an open-source prototype called Super-Agent, built to challenge Claude Code, Anthropic's coding tool 2. It watches developer group chats for bug reports and tries to reproduce and fix them without being asked, under a central coordinator that hands work to agents running in parallel; DeepSeek argues one person can otherwise oversee only three to five agent sessions at once 23.

In the team's tests, one task finished in just under 16 minutes against more than 29 the conventional way 3. DeepSeek itself calls the work an architectural exploration rather than a mature product, released on GitHub−1.38% — GitHub, down 1.38 percent today, and flags the security and privacy risks of agents wired into other systems 3.

The compute money behind the IPO wobbled

The market leg landed the same week. Microsoft cut its fiscal-year capital spending forecast to $170 billion from $175 billion, Amazon−1.20% — Amazon, down 1.20 percent today posted a second straight quarter of negative free cash flow, an outflow of $42.4 billion over 12 months, and Google's $9.1 billion outflow came with its worst trading day ever 4. Anthropic draws much of its compute capacity from Amazon, Google−0.82% — Google, down 0.82 percent today and SpaceX−5.26% — SpaceX, down 5.26 percent today; Microsoft, Google and Amazon are key partners and among its biggest investors 4.

The cloud giants Anthropic leans on just showed strain

  • Estimate
$0B$50B$100B$150B$200B$250BAmazon$220BAlphabet$205BUp to $205BMicrosoft$170BCut from $175B
Data
Value
Amazon (estimate)$220B
Alphabet (estimate)$205B
Microsoft (estimate)$170B
Fiscal-year capital spending guidance as stated in late-October 2026 results; all three are forecasts and Alphabet's is a ceiling. Source: CNBC, October 31, 2026.4

The growth story holds; the compute bill tests it

Demand is not the question: cloud growth beat estimates at all three companies, and analysts count 1,300% annual sales growth, 900% net revenue retention, meaning customers spend nine times what they spent a year ago, and an implied 85% gross margin in the prospectus 4. The wider IPO market has nonetheless been frozen since SpaceX's record June offering raised $86 billion 47.

PitchBook, the market-research firm, puts Anthropic's compute commitments near $855 billion, more than six times its $131.7 billion of June cash, and investors are still waiting on its third-quarter financials 874. On the usage counts alone, government reads as Anthropic's fastest-growing set of buyers, and from this week that set pays 40% less 1. What that does to the growth rate is the number the roadshow has to explain.

Get Model by email, free

The AI labs, their models, and the business behind them. In our daily email, every weekday.

Loading MSFT

Reader comments

0 comments

    Model

    Amazon hosts the Chinese AI model its own CEO is waiting to replace

    Andy Jassy put Z.ai's GLM 5.3 on Bedrock's six-to-eight-model shelf, called the platform gigantic, then said the West is hungry for a Western replacement. Z.AI's Singapore shares fell 8.62% the next day.

    Model

    Neurosurgeons' video model built on Alibaba's Wan lists at half a cent a second on fal and OpenRouter

    Two neurosurgeons shipped the cheapest image-to-video model on fal and OpenRouter at half a cent a second, claiming living-neuron research made an Alibaba model five times faster. The platform hosting it just had its founder beg publicly for scarce GPUs.

    Model

    Anthropic's $16.3 billion half-year meets MiniMax's $0.15 rate card

    MiniMax, named in Anthropic's distillation evidence, sells a model tied for fourth among open weights at about a fifteenth of Kimi's token price. Anthropic's Nasdaq debut, with no price range set, could trade as soon as November 16 and will set the multiple Chinese AI is graded on.

    Model

    Mistral publishes weights, halves prices, and makes resellers sign a deal; seven Chinese models outrank it

    Le Chonk's open weights arrive with a resale license, a memory wall and a scoreboard that still ranks seven Chinese models ahead. Sovereign buyers get AI no provider can switch off; Mistral gets a funnel into its own paid serving.

    Model

    DeepMind says coordination beats benchmarks while Alphabet posts negative $9.07 billion free cash flow

    DeepMind's new essay argues solo benchmarks measure the wrong thing and that human wages will anchor to the cost of compute. The same week Alphabet posted negative free cash flow, launched a number-one benchmark model developers cannot buy, and Anthropic headed toward a mid-November IPO.

    Model

    Nvidia made batch transcription free. Only the phone call still pays.

    Nvidia released a free, open-weights 40-language transcription model eight days after Mistral priced batch speech-to-text at fractions of a cent. Per-minute APIs still sell where latency forces streaming, and independent benchmarks keep undercutting the vendors' own clocks there.

    Get the newsletter

    The day's AI and market news, in your inbox every weekday.