Opening verdict
The headline question is usually framed as date, ticker and price. None was public by this note’s 12 August 2026 evidence cutoff. The more important question is whether a public record supports the economic case. It did not yet: Buteon found no public Anthropic S-1 or S-1/A in EDGAR after searches by company name, registrant wording and form type.
That absence does not contradict the company’s announcement. A confidential draft is reviewed nonpublicly. The SEC process allows an issuer to test and revise disclosure before making the registration statement and prior drafts public. Until that public filing appears, the issuer has information and investors have a narrative.
The disciplined conclusion is therefore asymmetric. Anthropic’s move toward an IPO is observed and its reported growth is substantial. Whether the economics can support the private valuation—or any eventual public value—remains untested from public financial disclosure.
What is confirmed—and what is not
- Confidential draft S-1
- Confirmed by Anthropic on 1 June 2026 for a proposed IPO of common stock.Company-confirmed
- Public S-1 or S-1/A
- Not found in Buteon’s EDGAR review through 12 August 2026.Buteon EDGAR search
- IPO date
- Not public. SEC review and market conditions do not create a fixed calendar.Observed
- Exchange and ticker
- Not public in company or SEC materials reviewed for this note.Observed
- Shares and offer price
- Not set when Anthropic announced the confidential submission.Company-confirmed
- Latest private valuation
- $965 billion post-money in the Series H announced 28 May 2026—not an IPO valuation.Company-reported
- Latest disclosed business scale
- Run-rate revenue crossed $47 billion earlier in May 2026, according to Anthropic—not audited annual revenue.Company-reported
- Public share access
- No publicly listed Anthropic common stock was available by the evidence cutoff.Observed
Funding, valuation and the limit of the headline numbers
Anthropic’s disclosed financing path changed scale rapidly. Three announcements moved the post-money valuation from $183 billion to $380 billion and then $965 billion in under nine months, alongside company-reported run-rate revenue moving from more than $5 billion to $14 billion and then more than $47 billion.
Each point is useful and each has a boundary. A post-money private-round valuation reflects the latest financing terms, not necessarily the price of every share class or a future public equity value. Run-rate annualizes a recent pace; it is not a substitute for revenue recognized under audited accounting over a completed year.
Private rounds to confidential review
Round terms and run-rate figures are reproduced from Anthropic’s announcements.
Company-confirmed Series F
- Transaction
- $13B raised
- Valuation
- $183B post-money
- Run-rate context
- More than $5B in Aug 2025
Anthropic said it served more than 300,000 business customers and would use the financing for capacity, safety research and international expansion.
Anthropic Series F announcement (opens in a new tab)Company-confirmed Series G
- Transaction
- $30B raised
- Valuation
- $380B post-money
- Run-rate context
- $14B company-reported
The round included part of previously announced Microsoft and NVIDIA investments, so it should not be treated automatically as entirely new capital on top of every earlier commitment.
Anthropic Series G announcement (opens in a new tab)Company-confirmed Series H
- Transaction
- $65B raised
- Valuation
- $965B post-money
- Run-rate context
- Crossed $47B earlier in May
Anthropic said the round included $15 billion of previously committed hyperscaler investments. The announcement linked the capital to research, compute, products and partnerships.
Anthropic Series H announcement (opens in a new tab)Company-confirmed Confidential draft S-1
- Transaction
- Not an offering
- Valuation
- No public offer value
- Run-rate context
- No public prospectus
The submission opened a nonpublic SEC review path. Anthropic said the number of shares and price had not been set and that any offering depended on market conditions and other factors.
Anthropic confidential draft announcement (opens in a new tab)
Series H valuation ÷ reported run-rate
$965B ÷ $47B. This is not a public-market multiple: the numerator is a private post-money round value and the denominator is a company-reported annualized pace, not audited trailing revenue.
Series F to Series H valuation step-up
$965B ÷ $183B across disclosed post-money values. Different rounds can carry different rights and do not reveal the fully diluted public-share economics.
What a confidential S-1 does—and does not—mean
A confidential draft gives SEC staff a nonpublic document to review. It lets an issuer receive comments and revise disclosure before public investors can read it. The company’s announcement establishes that Anthropic entered that process; it does not establish that the SEC has approved the offering, that a road show is scheduled, or that the issuer will proceed.
The SEC’s current procedures allow initial issuers to use nonpublic review if they publicly file the registration statement and the prior nonpublic drafts at least 15 days before a road show—or, without one, at least 15 days before the requested effective date. The public filing is the moment when investors can begin evaluating the actual prospectus rather than the issuer’s summary announcement.
Under SEC procedures, the registration statement and prior draft submissions must still be filed publicly before the offering can proceed. A confidential submission is not that public filing.
The public S-1 gaps that still matter
Form S-1 is valuable because it forces a connected record: audited financial statements beside management’s discussion, risk factors, ownership, related-party transactions, dilution, use of proceeds and the proposed distribution. Company updates can report growth; they cannot provide that whole reconciliation.
Buteon searched EDGAR through the evidence cutoff for public S-1 and S-1/A filings associated with Anthropic and found references to the company in other issuers’ filings, but no public Anthropic registration statement. “Not found” is a search result with a date boundary, not a claim about what the SEC may hold nonpublicly.
Audited financial history
Revenue, cost of revenue, operating expenses, losses or profit, balance sheets and cash flows on a consistent audited basis.
Revenue quality
Recognized revenue definitions, period history, product or geography mix, contract duration, pricing, credits and retention.
Compute economics
Large capacity agreements are announced; gross margin, utilization, depreciation, committed spend and unit economics are not publicly reconciled.
Concentration and dependencies
Cloud and chip partners are named; customer concentration, supplier exposure, minimum commitments and contractual dependencies are not quantified publicly.
Capital structure and ownership
Share classes, preferences, fully diluted ownership, option overhang, conversion, dilution and the economics held by strategic investors.
Governance and control
The PBC and Long-Term Benefit Trust framework is described; post-offering voting rights, charter terms and control mechanics are not in a public prospectus.
Material risks and contingencies
Issuer-specific litigation, regulation, safety, security, IP, data, export-control, talent, reliability and commercialization risks in S-1 form.
Offering terms
Exchange, ticker, shares, price range, primary versus secondary mix, underwriters, lockups, dilution and use of proceeds.
The public filing will not answer every question, but it is the first document designed to put these questions on one accountable record.
Anthropic would not be the first listed model company
The AI listing precedent already exists. HKEX identifies Zhipu AI and MiniMax as the first listed large-language-model companies; they began trading in Hong Kong on 8 and 9 January 2026, respectively. HKEX reported IPO proceeds of about $558 million for Zhipu and $711 million for MiniMax using Dealogic data.
That matters for accuracy and for comparison design. Anthropic would be a major US listing candidate, but it would not open public markets to model builders for the first time. Investors already have prospectuses and post-listing disclosure from two standalone model-platform issuers, alongside indirect exposure through public parents such as Alphabet, Meta, Alibaba and NTT.
The precedents are not valuation shortcuts. Hong Kong’s Chapter 18C regime, the issuers’ scale, product mix, geography, customer base, model-distribution strategy and capital requirements differ from Anthropic’s. They are most useful as disclosure precedents: examples of how a model company describes R&D, compute, commercialization and risk once it enters public markets.
A representative global model-lab market map
Anthropic’s public-market case sits inside several different corporate forms. Some labs are private standalone companies. Some are controlled programs inside listed parents. Two are already standalone public issuers. Their models reach users through managed applications, APIs, cloud platforms, downloadable weights or combinations of those channels.
The matrix is descriptive, not a ranking. It excludes benchmark scores and capability claims because those change quickly and do not establish financial quality. The relevant public-market questions are who owns the economics, who funds the compute, which channel controls distribution and whether a buyer receives direct or parent-company exposure.
Structure, distribution and public exposure
| Lab | Region | Structure | Distribution | Commercial position | Capital / reach | Public status |
|---|---|---|---|---|---|---|
| AnthropicCompany (opens in a new tab) · Series H (opens in a new tab) · Compute (opens in a new tab) | United States | Independent PBC with Long-Term Benefit Trust governance | Managed Claude products and API; AWS, Google Cloud and Microsoft Azure | Enterprise, developer and consumer subscriptions and usage | Private and strategic investors; large multi-cloud compute agreements | Private; confidential draft submitted; no public S-1 found |
| OpenAIStructure (opens in a new tab) | United States | OpenAI Group PBC controlled by the OpenAI Foundation | Managed applications and API, plus selected open-weight models | Consumer, developer and enterprise products | Private capital and strategic cloud distribution | Private; not separately listed |
| Google DeepMindAbout (opens in a new tab) · Models (opens in a new tab) | United States / United Kingdom | AI research and product organization within Alphabet | Managed Gemini products and APIs; Gemma open models | Consumer, cloud, workspace and research channels | Alphabet balance sheet and Google product distribution | Public-parent exposure; not a standalone listed lab |
| Meta AI / LlamaLlama (opens in a new tab) | United States | AI program within Meta Platforms | Downloadable Llama model weights through Meta and Hugging Face | Developer and commercial ecosystem around licensed model weights | Meta balance sheet and broad cloud and startup platform support | Public-parent exposure; not a standalone listed lab |
| DeepSeekDeepSeek-R1 (opens in a new tab) · Terms (opens in a new tab) | China | Products and services operated by Hangzhou DeepSeek Artificial Intelligence Co., Ltd. | Downloadable model weights plus hosted chat and API | Developer and consumer access with open deployment options | Open-weight repository plus hosted chat and API distribution | Standalone listing status not established by the cited first-party sources |
| QwenQwen3 (opens in a new tab) | China | Model program developed by the Qwen team at Alibaba Cloud | Open-weight releases plus managed cloud models and services | Developer, enterprise and Alibaba Cloud channels | Alibaba balance sheet and cloud distribution | Public-parent exposure; not a standalone listed lab |
| Zhipu AI / Z.aiGLM (opens in a new tab) · HKEX (opens in a new tab) | China | Standalone model-platform issuer | Open-weight GLM releases plus hosted chat and API | Enterprise and developer model platform | Public capital, enterprise relationships and developer distribution | HKEX-listed since 8 January 2026 |
| MiniMaxModels (opens in a new tab) · HKEX (opens in a new tab) | China | Standalone model-platform issuer | Open-weight models, managed APIs and consumer applications | Multimodal consumer, developer and enterprise products | Public capital and its own open platform | HKEX-listed since 9 January 2026 |
| Sakana AISeries B (opens in a new tab) | Japan | Independent private AI R&D company | Research, model orchestration and custom enterprise applications | Japanese enterprise and sovereign-AI focus | Venture and strategic capital plus Japanese enterprise partners | Private; not separately listed |
| NTT tsuzumiNTT (opens in a new tab) | Japan | Model and service program within NTT | Compact models through NTT Group, including on-premises and private cloud | Japanese business and government document workloads | NTT R&D, infrastructure and group distribution | Public-parent exposure; not a standalone listed lab |
Where the public-market case is strongest
First, the process has moved beyond rumor. Anthropic named the form, regulator and proposed security in a Rule 135 announcement. It also added public-company governance experience before the submission, including additional board appointments.
Second, the company’s own operating indicators show extraordinary scale. The disclosed run-rate moved from more than $5 billion in August 2025 to $14 billion in February 2026 and more than $47 billion in May. Anthropic also reported growth in large enterprise accounts and Claude Code revenue in the Series G announcement.
Third, Anthropic has raised capital and secured distribution at a scale few independent labs can match. Claude is distributed through its own products and the three largest cloud platforms. Anthropic says it trains and runs Claude across AWS Trainium, Google TPUs and NVIDIA GPUs rather than relying on one chip architecture.
Primary source: Anthropic’s Google and Broadcom compute announcement (opens in a new tab).
Finally, the governance design is more visible than at many private companies. Anthropic is a PBC, and the Long-Term Benefit Trust appoints directors and had appointed a majority of the board by April 2026. That is not the same as knowing post-IPO control, but it gives a public prospectus a concrete structure to explain.
Where the case remains weakest
The weak point is not demand storytelling. It is economic reconciliation. Run-rate can move faster than recognized revenue, especially in usage-based products with credits, contract ramps and volatile consumer demand. Without audited periods, investors cannot test seasonality, pricing durability, retention, revenue recognition or the share of growth funded by incentives.
Compute is simultaneously a moat, a supplier dependency and a cost obligation. Anthropic announced a commitment of more than $100 billion over ten years to AWS technologies, up to five gigawatts from Amazon and multiple gigawatts of next-generation TPU capacity through the Google and Broadcom arrangement. The public record does not yet connect those commitments to utilization, gross margin, depreciation, cash requirements or cancellation rights.
Primary sources: Anthropic’s Amazon compute announcement (opens in a new tab); Anthropic’s Google and Broadcom compute announcement (opens in a new tab).
The capital structure is also opaque. A $965 billion post-money valuation does not reveal preferences, conversion terms, fully diluted ownership or what portion of future proceeds would fund the company rather than selling holders. Strategic investors can be suppliers, distributors, customers and shareholders at once; the related economics require prospectus disclosure.
Governance needs the same treatment. The Long-Term Benefit Trust’s public mission and board role are material, but public buyers would need the actual charter, voting classes, director election rights, protective provisions and amendment mechanics after an offering. Mission governance can be a strength while still creating rights that common shareholders must understand.
Buteon view
Classify Anthropic as a confirmed confidential-IPO process with a still-unverified public investment case. The distinction is not semantic. Process evidence answers whether a company has started regulator review. Investment evidence answers what a public buyer would own, at what terms, with what financial history and risks.
The private financing record raises the burden of proof rather than lowering it. At $965 billion, small differences in revenue quality, margin, capital intensity or control can move the economic conclusion materially. The public S-1 should therefore be read first as a reconciliation document, not as a countdown marker.
The most useful future comparison will not be a single headline multiple against another lab. It will separate direct model-company issuers from public-parent programs, then normalize recognized revenue, gross margin, compute commitments, customer concentration, voting rights and primary versus secondary proceeds. Until those inputs are public, precision would be theatrical.
What to watch next
- Public S-1 or S-1/A: the first accountable view of audited financials, risk factors, ownership and the draft offering.
- Revenue reconciliation: reported run-rate versus recognized revenue across audited periods, including growth, mix, retention and pricing.
- Gross margin and compute: cost of revenue, depreciation, capacity commitments, utilization and how cloud-partner economics flow through the income statement and cash flow.
- Customer and supplier concentration: exposure to major enterprise accounts, hyperscalers, chip vendors and strategic counterparties.
- Capital structure: preferred-share conversion, fully diluted ownership, option overhang, voting classes and Long-Term Benefit Trust rights after listing.
- Offering construction: primary versus secondary shares, intended use of proceeds, dilution, underwriters and lockup arrangements.
- Actual terms: exchange, ticker, indicated price range and share count only when disclosed by the issuer in filed materials.
- Outcome: effectiveness, pricing and first trade—or a delay, withdrawal or revised transaction.
This note should be revised when a public filing changes the evidence boundary, not when an unattributed calendar date or synthetic quote page changes the rumor cycle.
Anthropic IPO FAQ
Is Anthropic publicly traded?
No. Anthropic remained privately held as of 12 August 2026. A confidential draft S-1 starts a possible IPO process; it does not create publicly listed shares.
Has Anthropic filed an S-1?
Anthropic confirmed that it confidentially submitted a draft Form S-1 on 1 June 2026. Buteon found no public Anthropic S-1 or S-1/A in EDGAR through 12 August 2026. SEC guidance distinguishes a confidential submission from a public registration-statement filing.
When is the Anthropic IPO date?
No IPO date was public by the evidence cutoff. The company said a proposed offering depends on SEC review, market conditions and other factors.
What will the Anthropic stock ticker be?
No ticker or exchange was public in the company and SEC materials reviewed for this note. Third-party quote pages do not establish an issuer-approved symbol.
What is Anthropic’s valuation?
Anthropic announced a $965 billion post-money private valuation with its Series H on 28 May 2026. That is the value attached to a private financing round, not an announced IPO valuation or public market capitalization.
Is Anthropic’s $47 billion run-rate audited revenue?
No public audit accompanies that figure. Anthropic said its run-rate revenue crossed $47 billion earlier in May 2026. Run-rate annualizes a recent pace and is not the same measure as revenue recognized over an audited fiscal year.
Can retail investors purchase Anthropic stock now?
Not through a public exchange as of the evidence cutoff. Private-market interests, where available, can carry different rights, restrictions, liquidity and eligibility and are not equivalent to listed common stock.
Would Anthropic be the first publicly listed AI model company?
No. HKEX identifies Zhipu AI and MiniMax as the first listed large-language-model companies; they listed in Hong Kong in January 2026. Anthropic would differ in scale, market, corporate structure and distribution.
Sources and methodology
- Anthropic — confidential draft S-1 announcement (opens in a new tab)
Primary confirmation of the confidential submission, proposed common-stock IPO and unset share and price terms.
- SEC — draft registration statement processing procedures (opens in a new tab)
Nonpublic review mechanics and the public-filing requirement before a road show or requested effectiveness.
- SEC — what is a registration statement? (opens in a new tab)
Prospectus, audited financial statements and the relationship between Form S-1, Regulation S-K and Regulation S-X.
- SEC — Form S-1 (opens in a new tab)
Primary form instructions for offering, business, financial, ownership, management and other disclosures.
- Anthropic — Series F (opens in a new tab)
$13 billion financing, $183 billion post-money value and company-reported August 2025 run-rate.
- Anthropic — Series G (opens in a new tab)
$30 billion financing, $380 billion post-money value, $14 billion reported run-rate and prior strategic-investment overlap.
- Anthropic — Series H (opens in a new tab)
$65 billion financing, $965 billion post-money value, reported $47 billion run-rate and hyperscaler-investment overlap.
- Anthropic — Amazon compute agreement (opens in a new tab)
AWS capacity, ten-year technology commitment, strategic investment and distribution disclosures.
- Anthropic — Google and Broadcom compute agreement (opens in a new tab)
Multiple gigawatts of next-generation TPU capacity, multi-hardware training and distribution across the three largest cloud platforms.
- Anthropic — Long-Term Benefit Trust board appointment (opens in a new tab)
Company description of PBC governance and the Trust-appointed board majority in April 2026.
- HKEX — AI IPOs drive a strong start to 2026 (opens in a new tab)
Listing dates, platform classification and proceeds for Zhipu AI and MiniMax.
- HKEX — specialist tech fundraising through Chapter 18C (opens in a new tab)
Exchange statement identifying Zhipu and MiniMax as the first listed large-language-model companies.
- OpenAI — corporate structure (opens in a new tab)
Foundation control of OpenAI Group PBC and the relationship between mission governance and equity.
- Google DeepMind — models (opens in a new tab)
Managed Gemini channels and open-model Gemma distribution.
- Meta — Llama distribution (opens in a new tab)
Current first-party model repository for downloadable Llama weights, licensing and distribution routes.
- DeepSeek — R1 repository (opens in a new tab)
First-party model repository, downloadable weights and licensing.
- DeepSeek — terms of use (opens in a new tab)
First-party identification of the legal operator of DeepSeek products and services; no funding or listing-status inference is drawn from it.
- Qwen — Qwen3 repository (opens in a new tab)
Alibaba Cloud team attribution, downloadable models and deployment routes.
- Z.ai — GLM repository (opens in a new tab)
Open model and hosted API distribution for the Zhipu/Z.ai platform.
- MiniMax — open-model and API announcement (opens in a new tab)
Downloadable weights, hosted API and consumer product distribution.
- Sakana AI — Series B (opens in a new tab)
Private funding, model-orchestration strategy and Japanese enterprise positioning.
- NTT — tsuzumi 2 update (opens in a new tab)
Compact model deployment through on-premises, private-cloud and NTT Group channels.
Disclaimer
Buteon is a research tool. This note is educational, uses public information reviewed through 12 August 2026, and does not provide investment advice or recommend purchasing, selling, subscribing for, or avoiding any security. Anthropic’s company-reported run-rate and private financing terms are not audited public-company results. Offering facts may change after publication. Nothing here is a valuation opinion or a prediction of whether or when an IPO will occur.