BusinessMarkets, money & strategy
Anthropic files S-1 for Nasdaq IPO, reporting $16.3B first-half revenue and first operating profit
Anthropic's public S-1 sets up a Nasdaq listing under the ticker ANT and shows revenue of $16.3B in the first half of 2026, a first operating profit in Q2, an $80.7B net loss driven largely by accounting charges, and extensive risk disclosures.
From the CNBC · Technology article of October 27, 2026 and cited sources: Revenue grew from $4.6B for all of 2025 to $16.3B in the first half of 2026. [2] [8] [6]
SOURCES[2] Yahoo Finance: Anthropic IPO live updates finance.yahoo.com[8] Yahoo Finance: Anthropic publicly files IPO prospectus finance.yahoo.com[6] The Next Web: Anthropic lost $42bn in 2025, Reuters reports thenextweb.com
AT Pulse research deskPublished · Updated · 5 min read · AI-assisted analysisAI-assisted, not human-edited. Methodology Corrections Who owns this site
- What happened
- Revenue grew from $4.6B for all of 2025 to $16.3B in the first half of 2026. [2] [8] [6]
- Original source
- CNBC · Technology article, independent perspective
- Brief by
- AT Pulse research desk AI-assisted; publisher and contact on the desk page
- When
- October 27, 2026 article published
- Covered
- Oct 27 edition
- Evidence
- 8 sources across 7 sites
- 1CNBC: Anthropic's IPO is risky businesscnbc.com
- 2Yahoo Finance: Anthropic IPO live updatesfinance.yahoo.com
- 3Malay Mail: Pentagon calls Anthropic a supply-chain riskmalaymail.com
- 4Digital Policy Alert: Commerce terminates Fable 5/Mythos 5…digitalpolicyalert.org
- 5Washington Technology: Judge rules designation illegal and…washingtontechnology.com
- 6The Next Web: Anthropic lost $42bn in 2025, Reuters reportsthenextweb.com
Reach: not measured yet
Revenue grew from $4.6B for all of 2025 to $16.3B in the first half of 2026. Q2 alone brought in $11.6B and produced a $974M operating profit. These figures come from the company's own filing, and no independent analysis of them has been published yet. Over the same period, accounting charges drove most of an $80.7B net loss. In 2025, about $34B of a $42B loss came from revaluing financing that could convert into shares. [2] [8] [6]
Compute is the main fixed cost. Quarterly compute spending rose from $3.4B in Q1 to $5.6B in Q2 and is projected at about $3.1B per month once full capacity arrives in December 2026. Reported infrastructure commitments run from $430B locked in to as much as $637B including cancellable deals. The earlier confidential draft put the figure at $518B, about 80% of it non-cancellable. Anthropic is also building its own data centers to rely less on AWS and Google Cloud. [2] [8] [6]
The risk factors run past 90 pages. They warn that Anthropic's products could pose catastrophic or existential risks and that liability from a major incident could exceed what insurers can cover. They also cover data-center moratoriums under consideration in 16 states, minors reaching the models despite the terms-of-service ban, and customers switching to cheaper or open-source models. [1] [2] [8]
Government actions have already interrupted access to Anthropic's models. In February 2026 the Pentagon labeled Anthropic a supply-chain risk, and a federal judge ruled that designation unlawful in August. In June, a Commerce Department order suspended Claude Fable 5 and Mythos 5. It was lifted after Anthropic agreed to coordinate with the government on security, and Commerce can reimpose it if Anthropic does not keep those commitments. [3] [4] [5]
The filing shows fast revenue growth alongside hundreds of billions in mostly fixed compute commitments.
Read the full assessment
Together with past government actions that cut off model access, this suggests pricing and availability risk for organizations that depend on Claude.
Executive brief
Anthropic is asking public-market investors to fund a company whose prospectus says its own products could pose "catastrophic or existential risks to humanity." The filing also shows profitability arriving at the same time as very large losses. Revenue reached $16.3B in the first half of 2026, up from $4.6B for all of 2025, and Q2 produced the company's first operating profit ($974M). Net losses over the same period were $80.7B, mostly from accounting charges (Yahoo Finance live). The risk factors run past 90 pages and cover data-center backlash, power shortages, minors' access and a dispute with the Trump administration (CNBC).
What changed and event timeline
Pentagon labels Anthropic a supply-chain risk
This followed Anthropic's refusal to allow its models to be used for mass surveillance or autonomous weapons. President Trump ordered agencies to phase out its products over six months ().
Export controls imposed, then lifted
A Commerce order suspended Claude Fable 5 and Mythos 5. It was terminated on June 30 after Anthropic made security-coordination commitments ().
Court rules the designation unlawful
Judge Rita Lin (N.D. Cal.) found the Pentagon's measures "illegal and baseless," citing First and Fifth Amendment violations ().
Reuters reports on the confidential draft
The draft showed a $42B net loss for 2025 and at least $518B in infrastructure commitments, about 80% of them non-cancellable ().
Public S-1 filed
Anthropic plans to list Class A shares on the Nasdaq as "ANT." Goldman Sachs and Morgan Stanley lead the deal. Share count and price range are not yet set ().
CNBC analyzes the risk factors
Its analysis highlights data-center opposition, power shortages, child safety and the risk that customers adopt AI more slowly than expected ().
Capabilities and access
- Named models: Claude Fable 5 and Claude Mythos 5.
- Commerce kept the right to reimpose license requirements if Anthropic does not keep its commitments (same source).
- Minors are barred from Claude under its terms of service.
Read the full section
- Named models: Claude Fable 5 and Claude Mythos 5. During the June controls, Mythos 5 was first released only to some "trusted" US organizations (Digital Policy Alert).
- Commerce kept the right to reimpose license requirements if Anthropic does not keep its commitments (same source).
- Minors are barred from Claude under its terms of service. The filing nonetheless expects minors and other vulnerable users to reach the models as adoption grows (CNBC).
Technical analysis for researchers and developers
The prospectus is a financial document, not a model card. It discloses no architecture or benchmark results.
Read the full section
The prospectus is a financial document, not a model card. It discloses no architecture or benchmark results. What it does document:
- Compute dependence. Q2 compute costs were $5.6B, up from $3.4B in Q1, and are projected at about $3.1B per month once full capacity arrives in December 2026 (Yahoo Finance live).
- Anthropic Direct. The company is building its own data centers to reduce reliance on AWS and Google Cloud (same source).
- Limits of safety work. The filing says no safety approach can foresee every scenario, and that models may develop "unanticipated capabilities" (Yahoo Finance).
- Government coordination on releases. The export-control settlement ties future model releases to security-risk detection and coordination with the US government (Digital Policy Alert).
Claims and evidence
- Revenue and profit: $16.3B in H1 2026, with Q2 at $11.6B and a $974M operating profit.
- The $80.7B loss comes mostly from accounting charges (Yahoo Finance).
- Compute commitments are reported differently:
Read the full section
- Revenue and profit: $16.3B in H1 2026, with Q2 at $11.6B and a $974M operating profit. These are company-reported figures in an audited filing, but no independent analysis has been published yet (Yahoo Finance live).
- The $80.7B loss comes mostly from accounting charges (Yahoo Finance). In 2025, about $34B of the loss came from revaluing financing that could convert into shares (The Next Web).
- Compute commitments are reported differently:
- $430B locked in, up to $637B including cancellable deals (Yahoo Finance)
- $518B over a decade, per the confidential draft (The Next Web)
- Customer concentration is also reported differently: about one-fifth of revenue from two customers (Yahoo Finance) versus nearly 25% in 2025 (The Next Web).
Context and prior work
- Valuation: Anthropic was valued at $965B in May 2026. The IPO targets $1.5–2T. OpenAI was valued at $852B in March (Yahoo Finance live).
- Governance: Seven co-founders control 50.1% of voting power through a Founder LLC (The Next Web).
- Peer legal exposure: Meta's roughly $17B settlement with state attorneys general over child mental health and tech addiction sets the bar for youth-harm litigation that Anthropic says it could face (CNBC).
Read the full section
- Valuation: Anthropic was valued at $965B in May 2026. The IPO targets $1.5–2T. OpenAI was valued at $852B in March (Yahoo Finance live).
- Governance: Seven co-founders control 50.1% of voting power through a Founder LLC (The Next Web).
- Peer legal exposure: Meta's roughly $17B settlement with state attorneys general over child mental health and tech addiction sets the bar for youth-harm litigation that Anthropic says it could face (CNBC).
Limitations, safety and contested findings
- Catastrophic risk: The filing warns that liability from a catastrophic AI incident could exceed what the insurance industry can cover, and that tighter AI regulation could limit deployment (Yahoo Finance live).
- Data centers: Lawmakers in 16 states are considering data-center moratoriums. The filing says Anthropic's public profile may draw extra opposition to its projects (CNBC).
- Pentagon dispute: The government's case rested on calling Anthropic a national-security risk. The court rejected that basis, finding no meaningful threat (Washington Technology).
Read the full section
- Catastrophic risk: The filing warns that liability from a catastrophic AI incident could exceed what the insurance industry can cover, and that tighter AI regulation could limit deployment (Yahoo Finance live).
- Data centers: Lawmakers in 16 states are considering data-center moratoriums. The filing says Anthropic's public profile may draw extra opposition to its projects (CNBC).
- Pentagon dispute: The government's case rested on calling Anthropic a national-security risk. The court rejected that basis, finding no meaningful threat (Washington Technology).
Business and practitioner implications
- Cheaper models: Anthropic itself says customers may decide frontier performance gains aren't worth the extra cost and switch to cheaper or open-source models (CNBC).
- Fixed costs: About 80% of commitments are non-cancellable (The Next Web). Pricing will depend on revenue continuing to outpace those costs.
- Policy exposure: Export-control and procurement actions have interrupted access to Anthropic's models before. Government and regulated customers should plan for disruption.
Read the full section
- Cheaper models: Anthropic itself says customers may decide frontier performance gains aren't worth the extra cost and switch to cheaper or open-source models (CNBC). Teams relying on Claude should benchmark alternatives.
- Fixed costs: About 80% of commitments are non-cancellable (The Next Web). Pricing will depend on revenue continuing to outpace those costs.
- Policy exposure: Export-control and procurement actions have interrupted access to Anthropic's models before. Government and regulated customers should plan for disruption.