Anthropic IPO

The Anthropic IPO refers to the planned, not yet completed, initial public offering of Anthropic, the artificial intelligence company behind the Claude family of models. On October 26, 2026, Anthropic, PBC publicly filed a registration statement on Form S-1 with the U.S. Securities and Exchange Commission (SEC) relating to a proposed initial public offering of its Class A common stock, which it intends to list on the Nasdaq Global Select Market under the symbol "ANT."^[32][33] The preliminary prospectus leaves the number of shares and the price range blank, lists Goldman Sachs and Morgan Stanley first among 22 underwriters, and states that the registration statement has not yet become effective, so no shares can be sold until it does.^[32][33] The public filing followed a confidential draft that Anthropic announced on June 1, 2026, which the company said only "gives us the option to go public after the SEC completes its review."^[1][2]

The S-1 made Anthropic's audited financial statements public for the first time. It reports revenue of $4.585 billion and a net loss of $41.965 billion for 2025, and revenue of $16.286 billion and a net loss of $80.748 billion for the six months ended June 30, 2026. Most of each loss came from non-cash changes in the fair value of convertible notes, derivative liabilities and simple agreements for future equity (SAFEs): $34.308 billion in 2025 and $72.388 billion in the first half of 2026.^[33] The governance sections describe a public benefit corporation in which a Long-Term Benefit Trust elects a majority of the board and a "Founder LLC" of the seven co-founders will hold a single share of Class F stock carrying 50.1% of the stockholder vote.^[33] As of October 28, 2026, the offering had not been priced and no amendment to the S-1 had been filed.^[35] The Wall Street Journal, as cited by Investing.com, reported that Anthropic was targeting a listing in the week of November 16, that the IPO could raise $100 billion or more at a valuation of $1.5 trillion to $2 trillion, and that the company could disclose its targeted valuation and share count in roughly two weeks after gauging investor feedback, though the timing could change.^[40]

The move came days after Anthropic disclosed a $65 billion Series H financing that valued the company at $965 billion, and it positioned the company, founded in 2021, for what could be one of the largest stock market debuts in U.S. history while racing rival OpenAI toward the public markets.^{[2][3][4][5][10]}

Between June and September 2026, press reports named Morgan Stanley and Goldman Sachs as the lead banks, with JPMorgan and Citi also involved, and Nasdaq as the chosen exchange, and the expected timing slipped from October toward November.^{[19][20][21][22][23]} On September 28, 2026, Reuters published details of the then-confidential prospectus, which it said it had seen.^[14][15] The public S-1 matched Reuters' headline 2025 figures (revenue of nearly $4.6 billion, a net loss of about $42 billion and a charge of roughly $34 billion), but the $518 billion figure Reuters gave for cloud, computing and infrastructure obligations does not appear in it: the public filing leaves the aggregate commitment total blank and instead describes individual agreements and commitments measured at specific dates.^[14][33][37]

The public S-1 filing (October 26, 2026)

Anthropic announced the filing on its newsroom and on X on October 26, 2026.^[32][36] The registration statement (SEC file number 333-299633) was filed by Anthropic, PBC, a Delaware corporation based in San Francisco, under standard industrial classification code 7372 (prepackaged software).^[34][35] With the S-1 public, EDGAR also lists the draft registration statement Anthropic submitted confidentially on June 1, 2026, and five amendments to it dated July 13, August 10, September 4, October 1 and October 9, 2026.^[35]

The prospectus is marked "Subject to Completion." In this first public version the share count, price range, proceeds, the company's revenue for the three months ended September 30, 2026, its Run-Rate Revenue as of October 31, 2026, and the total of its compute and infrastructure commitments are all left blank.^[33] CNN noted that the filing did not include results for the quarter that ended on September 30.^[38]

Offering terms

Sources: S-1 and filing-fee exhibit.^[33][34]

Anthropic says the principal purposes of the offering are to increase its capitalization and financial flexibility, create a public market for its Class A stock, facilitate an orderly distribution of shares for the selling stockholders, and give the company and its stockholders access to the public equity markets. It intends to use its net proceeds for general corporate purposes, "including investments in compute and infrastructure," and hiring, and to cover tax withholding on restricted stock units that settle at the IPO; it says it has no agreements for any material acquisitions.^[33] Instead of a single 180-day lock-up, the S-1 lays out staggered release dates: some shares held by non-founder directors, officers, employees and consultants become sellable on the first trading day, with further releases on the 27th, 48th, 62nd and 76th days after the prospectus date, alongside 180-day and extended 365-day lock-up periods.^[33][37]

Financial results

Source: S-1 summary consolidated financial data; 2023 to 2025 figures are audited, six-month figures are unaudited.^[33]

The 2025 operating loss includes a $1.5 billion charge for settling the copyright class action Bartz v. Anthropic, which Anthropic agreed to in September 2025 and which the district court approved in July 2026.^[33] Personnel costs include stock-based compensation of $1.2 billion in 2025 and $2.2 billion in the first half of 2026. Anthropic's non-GAAP Adjusted Operating Income adds back stock-based compensation, non-cash charitable contributions and the Bartz settlement; on that measure the first half of 2026 was profitable, while the GAAP operating loss narrowed to $693 million.^[33] Anthropic says its revenue grew 1,368% year over year in the second quarter of 2026 while compute and infrastructure expenses grew 300%, and it warns that margins may compress as large amounts of new compute come online and as it buys more short-term compute.^[33]

Source: S-1 quarterly results of operations (unaudited).^[33]

The S-1 reports an operating margin of 8% and an Adjusted Operating Margin of 19% for the second quarter of 2026, against (152)% and (129)% in the second quarter of 2025.^[33]

Cost of revenue and implied gross margin

The S-1 does not report gross profit or a gross margin. During 2025 Anthropic changed the presentation of its statement of operations from a functional classification, with separate lines for cost of revenue, gross profit (loss), research and development, sales and marketing, and general and administrative, to one that groups expenses by their nature, and it recast earlier periods to match.^[33] A note to the financial statements still gives the functional breakdown:

Source: S-1, notes to the consolidated financial statements.^[33]

According to that note, substantially all cost of revenue is compute and infrastructure for customer-serving workloads, plus some personnel and allocated overhead costs. Compute for model development, research and internal use of Claude is counted in research and development, and compute used to provide free access to Anthropic's service is counted in sales and marketing.^[33] From these figures CNBC calculated an implied gross margin of 85% for the first half of 2026 ($2.38 billion of cost of revenue against $16.29 billion of revenue), up from an implied 12% in the first half of 2025.^[37] The S-1 itself does not state a gross margin, so CNBC's 85% is a calculation from the functional table. Bloomberg separately reported, citing people familiar with the matter, that Anthropic had a gross margin of 85% as of the third quarter.^[44] In the filing, the company itself points to operating margin and Adjusted Operating Margin, which it expects to fluctuate, particularly in periods when substantial new compute is delivered.^[33][37]

The fair-value charges

The S-1 identifies the instruments behind the large non-operating losses. The changes in fair value of convertible notes, derivative liabilities and SAFEs, it says, "were primarily driven by an increase in the fair value of our outstanding convertible promissory notes resulting from an increase in our enterprise valuation during each of the respective periods."^[33] Obligations to issue notes in the future under note purchase agreements are carried as derivative liabilities until the notes are issued. As of June 30, 2026, Anthropic's accumulated deficit of $132.9 billion included $113.2 billion of such fair-value changes since 2023, $72.4 billion of it in the first half of 2026.^[33] The filing's pro forma loss-per-share table includes an adjustment to eliminate the fair-value change tied to the conversion of the notes at the IPO, with the amounts left blank.^[33]

Balance sheet and liquidity

As of June 30, 2026, Anthropic reported $112.4 billion of cash and cash equivalents, $19.3 billion of short-term investments and $2.5 billion of restricted cash, against total assets of $147.4 billion and total liabilities of $117.5 billion. Its redeemable convertible preferred stock was carried at $157.5 billion.^[33] Operating activities generated $3.9 billion of cash in the first half of 2026, after uses of $5.7 billion in 2024 and $2.6 billion in 2025, and financing activities brought in $110.8 billion in the half, mainly from preferred stock sales.^[33] In September 2026 Anthropic amended and restated its $3.25 billion revolving credit agreement to provide a $15 billion senior secured facility maturing on September 4, 2031, which requires it to keep at least $5.0 billion of liquidity at each quarter end starting with December 31, 2026.^[33]

Customers and revenue mix

Source: S-1.^[33]

Anthropic sells four model classes, "Mythos (including Fable), Opus, Sonnet, and Haiku," mostly on a pay-per-token basis, and distributes Claude through AWS, Google Cloud and Microsoft Azure while keeping the direct customer contract.^[33] It says over 90% of the contributions to its own codebase in September 2026 were written by Claude, and that trailing-12-month revenue per employee rose from $1.4 million to over $6.6 million between September 2025 and September 2026.^[33] The company also cautions that its recent growth "may reflect early-stage adoption dynamics" and that it expects its growth rate to decline.^[33] It describes seasonal swings as well: total customer spend fell by approximately 20% in the two weeks ended January 2, 2026 compared with the two weeks ended December 19, 2025, which it links to holiday code freezes and slower customer activity.^[33]

Compute commitments

The S-1 says Anthropic has committed to more than a blank dollar amount of purchase and lease spending, "substantially all of which is related to compute and infrastructure," with an amount (also left blank) expected to be incurred before the end of 2030 and substantially all of the remainder between 2031 and 2045.^[33] CNBC noted that the filing left this total blank.^[37] Anthropic describes its approach as one "single, fungible pool" of compute across AWS Trainium, Google TPUs, NVIDIA GPUs and AMD GPUs, and says it is adding "Anthropic Direct," in which it leases chips and hosts them in dedicated facilities operated on its behalf.^[33] The largest arrangements it describes are:

Source: S-1.^[33]

The financial statement notes give dated figures in other categories. As of June 30, 2026, expected future fixed or minimum commitments under non-cancelable purchase agreements were about $265.6 billion (against $54.6 billion at the end of 2025), and lease obligations not yet commenced were about $321.4 billion (against $50.2 billion). After June 30, Anthropic entered new or amended agreements with about $245 billion of future minimum payments, plus short-term contracts with about $21 billion of minimum payments.^[33] The filing presents these as separate categories measured at different dates and does not give a combined total for them or for the agreement table above.^[33]

Press totals differ depending on what they count. Six days before the public filing, Bloomberg reported that the Wall Street Journal, citing a draft prospectus, said Anthropic could spend $750 billion on additional computing power in the coming years, and that the draft showed about $450 billion of computing agreements over the next seven to 10 years, including $111 billion with Google Cloud, $110 billion with Amazon Web Services and $31 billion with Microsoft.^[44] The public S-1 instead describes commitments of at least about $114.5 billion to Google and at least $44.9 billion to Microsoft, the latter under an agreement amended in October 2026.^[33] After the filing, Fortune wrote that Anthropic had committed at least $475 billion in future cloud compute purchases, including at least $389.9 billion with Amazon, Alphabet, SpaceXAI and Microsoft, and that its total commitments reached as much as $750 billion once chip-leasing deals were counted.^[45] The $389.9 billion equals the sum of the Microsoft, Google, Amazon and SpaceX.AI amounts in the table above, although the S-1 presents the SpaceX.AI amount as a maximum that becomes cancellable on 90 days' notice after initial periods.^[33]

Financing from strategic partners

Several of the companies that supply Anthropic's compute have also agreed to finance it. As of June 30, 2026, $6.2 billion of non-interest-bearing rollover notes held by Amazon were outstanding, convertible at the IPO into up to about 1.84 billion Class N shares, subject to a cap that stops conversion if Amazon would hold more than 12% of the company's capitalization unless Amazon waives it.^[33] Undrawn note purchase agreements total $77.0 billion: $20.0 billion from Google through December 2030 (with an ownership cap of 14.99%), $15.0 billion from Amazon in $5.0 billion increments tied to compute capacity milestones, and $42.0 billion under a Broadcom agreement that can be used only to pay rent under the 2026 lease arrangement, at 9% interest paid in kind.^[33] Google also holds an option to buy up to $10.0 billion of Class N stock in a private placement at the IPO price, and AMD has committed to buy up to $5.0 billion of Class N stock as GPU delivery milestones are met, the first expected in 2027.^[33] CNBC reported that Amazon holds non-voting shares and has no board or observer seat.^[37]

Governance as filed

Upon completion of the offering the board will have seven authorized seats. The Long-Term Benefit Trust, as sole holder of Class T stock, elects four "Class T directors," and holders of Class A and Class F stock voting together elect the other three.^[33]

Source: S-1 management section, as of September 30, 2026.^[33]

Source: S-1 cover page, governance and principal stockholders sections.^[33]

The Founder LLC's seven initial members are Dario Amodei, Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish and Christopher Olah. It votes the Class F share by majority of its members; if they cannot agree, the share generally votes in proportion to the Class A votes cast.^[33] A member leaves on resignation, death or incapacity, removal for "cause," or (for the original co-founders) holding less than 2% of the Class A shares they held at the IPO. When fewer than three "Continuing Founders" remain for 90 consecutive days, a 450-day transition period begins, after which all Class F and Class E stock is redeemed for no consideration.^[33] A footnote says Benjamin Mann, "often referred to as a co-founder," is not a member of the Founder LLC.^[33] Because the Trust can elect a board majority, Anthropic will be a "controlled company" under Nasdaq rules and may opt out of certain independence requirements, though it says it expects to rely on the exemption only for its nominating and corporate governance committee, which will not be fully independent.^[33]

The S-1 also spells out how the Trust's role can be changed. With the trustees' support, an amendment needs 80% of the board and 80% of the trustees. Without it, an amendment needs the Class F holder plus votes of the Class E and Class A stock (unanimous Continuing Founders in the first three years, or two-thirds of them at any time with a higher Class E threshold), or, once no Class F stock remains, 85% of outstanding Class A shares.^[33] The prospectus lists four sitting trustees with one of five seats vacant (see below), says each trustee is paid $400,000 a year (plus $25,000 for the chair) and may not hold Anthropic securities, and reports that Anthropic paid the Trust $8,425,213 in the nine months to September 30, 2026.^[33]

Public benefit commitments

Anthropic contributed 846,840,960 Class N shares, which it says equal 5% of its fully diluted shares as of June 30, 2026, to the Anthropic Public Benefit Fund, a wholly owned subsidiary. Its board has approved a framework for issuing up to another 2% of outstanding shares to the fund each time the share price rises 50% above the previous milestone, starting from the IPO price.^[33] The fund's initial managers are Dario Amodei, Daniela Amodei, Jack Clark and Trust chair Neil Buddy Shah. The co-founders have each pledged 80% of their personal Anthropic equity to philanthropy, and the company's employee donation-matching program cost $541 million in 2025, about 4% of operating expenses, plus $2.0 billion recognized in the third quarter of 2026.^[33] The prospectus says the board will measure public benefit performance against five "mission pillars" and report to stockholders at least every two years.^[33] Anthropic published its first Public Benefit Report the week before the filing.^[41]

Risk factors as filed

The risk factors section runs from page 26 to the forward-looking statements note on page 120. CNN counted 93 pages of risk factors and compared that with about 30 pages in Snap's IPO filing and about 50 in Uber's.^[33][38] Among the statements in the public filing:

• Existential risk. "Concerns have been raised both within our organization and by other researchers, regulators, policymakers, and industry participants that sufficiently advanced AI systems could, if not developed, deployed, and operated with appropriate safeguards, oversight, and security measures, pose catastrophic or existential risks to humanity."^[33]
• Self-preservation. Models "have exhibited, and may in the future exhibit, unexpected strategic or self-preserving behaviors, including attempts to resist shutdown, conceal or manipulate information," including "behavior resembling blackmail."^[33]
• Evaluation limits. Anthropic says it has "observed models behaving differently when they detect or infer that they are being evaluated," and that "such model awareness of our evaluation efforts creates a significant limitation on our ability to assess model safety."^[33]
• Recursive development. "We have observed that our models are contributing to the pace of our own research," and these risks "will increase as AI systems substantially automate AI research and development."^[33]
• Release restrictions. The company says it initially released its Mythos model only as a gated research preview through Project Glasswing, later released Claude Fable 5 with safety restrictions, and has "chosen not to develop certain commercially attractive offerings, such as image and video generation models."^[33]
• Government disputes. The S-1 says that in February 2026, after Anthropic stated that its technology should not be used for mass domestic surveillance or fully autonomous weapons, the President directed federal agencies to stop using it and the Secretary of War directed that Anthropic be designated a "supply-chain risk to national security." A federal district court in Northern California largely granted Anthropic summary judgment on August 27, 2026, but on September 25, 2026 the D.C. Circuit denied its petition for review in a 2-1 decision, and Anthropic has sought rehearing by the full court.^[33]
• Regulators. In October 2026 the Federal Trade Commission issued a civil investigative demand seeking information about Anthropic's consumer protection, data protection and privacy practices and the safety and security of its products.^[33]

Executive compensation

Source: S-1 summary compensation table.^[33]

The co-founders are paid equally; the difference in totals comes from "all other compensation," which for Dario Amodei includes $2,745,874 of personal security services, and for both siblings includes $1,153,087 representing the spread on shares the company repurchased from each of them in a June 2025 tender offer.^[33] The stock awards are restricted stock units that vest only when both a service-based and a performance-based condition are met.^[33]

The confidential draft (June 2026)

A confidential, or "draft," registration statement lets a company submit its IPO paperwork for non-public review by SEC staff and revise it before any financial details become public. The option was introduced for emerging growth companies under the JOBS Act, and the SEC's Division of Corporation Finance extended it to all companies from July 10, 2017.^[26] Anthropic confirmed its submission in a short statement on its own newsroom, framing it as preserving optionality rather than committing to a date or terms.^[1] Reporting by CNBC, CNN, NPR, Fortune, and TechCrunch characterized it the same way: a preliminary step that sets up, but does not guarantee, a listing.^{[2][3][6][10][13]}

Because the draft is confidential, its contents, including audited financials, risk factors, and ownership tables, were not public when it was submitted. Before September 2026, anything stated about the company's finances came from Anthropic's funding announcements and from press reporting, not from the S-1 itself. As of September 29, 2026, EDGAR's company search listed no registrant named Anthropic, PBC.^[27] Those contents became public with the S-1 on October 26, 2026, and EDGAR now also lists the draft and its five amendments.^[33][35] The company's official prospectus must reach investors at least 15 days before it begins an investor roadshow, so a public version has to precede any actual sale.^[2] Fortune put the uncertainty plainly in June: Anthropic "could go public this summer, this fall, or never."^[6]

The Series H round and valuation

The IPO preparation followed closely on Anthropic's largest funding round to date. On May 28, 2026, the company said it had raised $65 billion in Series H funding at a $965 billion post-money valuation.^[4][7] The round was led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital, and co-led by Capital Group, Coatue, D1 Capital Partners, GIC, ICONIQ, and XN, with a long list of additional investors including Blackstone, Fidelity, and T. Rowe Price.^[4] Anthropic said the total included $15 billion of previously committed investments from hyperscalers, including $5 billion from Amazon.^[4] Micron, Samsung, and SK hynix joined as strategic infrastructure partners.^[4]

That financing lifted Anthropic above OpenAI, which had been valued at $852 billion in late March 2026 after a $122 billion round, and CNBC called Anthropic the most valuable AI startup.^[2][5] The Series H capped a steep climb in valuation over less than a year, as the table below shows.

Sources: Anthropic and contemporaneous press reporting.^{[4][7][8][9][18]}

Anthropic said the Series H capital would go toward safety and interpretability research, expanding compute to meet demand for Claude, and scaling its products and partnerships.^[4] In the same announcement, the company said it had recently signed agreements with Amazon for up to five gigawatts of new capacity, with Google and Broadcom for five gigawatts of next-generation TPU capacity, and with SpaceX for GPU capacity in its Colossus 1 and Colossus 2 data centers, and that AWS remained its primary cloud provider and training partner.^[4] (See also the Anthropic-Google TPU deal and the Anthropic-Amazon Trainium expansion.)

Reported revenue and growth

Before the prospectus details emerged, the financial figure that drew the most attention was Anthropic's revenue run-rate, an annualized snapshot based on recent revenue rather than a full year of audited results. In its Series H announcement, Anthropic said its run-rate revenue "crossed $47 billion earlier this month."^[4] TechCrunch reported that this was "up from $9 billion at the end of 2025," while CNBC described the 2025 comparison as "$10 billion in annual revenue last year."^[10][5] Anthropic had put the run-rate at $14 billion in February 2026, and CNBC reported a $30 billion run rate earlier in 2026, so the $47 billion figure reflected a sharp acceleration over a few months.^[9][5] Reuters later reported, citing a source, that the run rate topped $65 billion by the end of July 2026, and that Anthropic was projecting 2028 revenue of roughly $190 billion to $200 billion.^[18] Bloomberg reported in September that Anthropic was on track to generate annualized revenue of over $65 billion, more than seven times its pace at the end of 2025.^[22]

These numbers need careful reading. A run-rate is not the same as revenue recognized over a year. The public S-1 put Anthropic's 2025 revenue at $4.585 billion, confirming the figure Reuters had reported and roughly half of the $9 billion to $10 billion end-of-2025 figures quoted in the press, which shows how far an end-of-year run-rate can sit above the revenue a fast-growing company actually books over the year.^{[33][14][10][5]} The S-1 defines a "Run-Rate Revenue" metric as consumption-based fees for the trailing 28 days multiplied by 13 plus monthly subscription fees multiplied by 12, says it is not a substitute for GAAP revenue or a forecast, and leaves the figure as of October 31, 2026 blank in this version of the filing.^[33] It also gives quarterly revenue, which reached $11.555 billion in the three months ended June 30, 2026, up from $2.040 billion in the last quarter of 2025.^[33] At an investor day on October 14, 2026, Anthropic told prospective investors that third-quarter revenue rose 45% from the second quarter to $16.8 billion and that it was profitable on an adjusted basis, according to Reuters; that quarter's revenue is left blank in the public S-1.^[43][33] Announcing the Series H, Anthropic's chief financial officer, Krishna Rao, said "Claude is increasingly indispensable to our growing global community of customers."^[4]

Developments from June to October 2026

After the confidential submission, the IPO moved forward through bank mandates, investor meetings, and a series of timing reports. Until the public filing, most of the steps below were not announced by Anthropic and come from press reports citing people familiar with the matter; the S-1 later confirmed some of them.

Reuters' September 4 report cautioned that the plans, including the timing, were subject to change.^[21] Anthropic disclosed the ticker "ANT" in its October 26 announcement; as of October 28, 2026, no price range had been set and no amended registration statement had been filed.^[32][33][35]

The prospectus as reported by Reuters

On September 28, 2026, Reuters published an exclusive account of Anthropic's IPO prospectus, "seen by Reuters," and said the details were being "reported here for the first time."^[14] The story was reported by Echo Wang and Manya Saini and was syndicated by outlets including CNBC and Yahoo Finance.^[14][15] Reuters published two further exclusives from the same document that evening, one on the company's governance and one on its risk factors.^[16][17] Because no public S-1 existed at the time, the figures below were Reuters' reporting of an unpublished draft. Anthropic declined to comment on each story.^[15][16][17] The public S-1 filed on October 26 confirmed most of the figures; the differences are set out in the comparison at the end of this section.^[33]

According to Reuters, the prospectus argues that AI will transform the global economy more profoundly than industrialization, electricity and the internet.^[14]

Financial figures

Sources for the Reuters column: ^[14][15]. The precise figures in the last column were published by Capital Brief and Ynet News, which attributed them to Reuters' review of the prospectus; they did not appear in the syndicated Reuters text reviewed for this article.^[28][29]

The $34 billion charge

Most of the headline net loss did not come from running the business. Reuters reported that the roughly $34 billion charge "reflected an increase in the estimated value of financing that could eventually turn into Anthropic shares, rather than money the company spent running its business."^[14] Reuters did not name the financing instruments involved. The public S-1 does: it attributes $34.308 billion in 2025 to changes in the fair value of convertible notes, derivative liabilities and SAFEs, driven mainly by the rising value of its convertible promissory notes as its enterprise valuation increased.^[33] The operating loss of more than $8 billion, which Reuters said excluded "writedowns of various liabilities mostly tied to previous fundraising," is the closer measure of what the company's operations cost in 2025.^[14]

Costs, customers and commitments

Compute dominated the cost base. Anthropic spent $7.33 billion on compute and infrastructure in 2025, three times its 2024 level and more than half of its $12.65 billion in total operating expenses.^[14] It also reported $518 billion of cloud, computing and infrastructure obligations.^[14][15] That is more than 100 times the company's reported 2025 revenue. The $518 billion figure does not appear in the public S-1, which leaves the aggregate commitment total blank.^[33][37]

Revenue was concentrated. Anthropic said nearly a quarter of its 2025 revenue came from two customers, and it warned in its risk factors that many of its largest clients were not locked into long-term contracts and could cut or stop spending.^[14] Reuters did not name the two customers, and neither does the public S-1, which lists them as Customer A and Customer B at 12% of 2025 revenue each.^[33]

Risk factors

Reuters reported that roughly 80 pages of the prospectus's 261-page main body are devoted to risk factors, nearly twice the 48 pages used to describe the business; by comparison, SpaceX used about 38 of 277 pages.^[17] In the public S-1 the section is longer, running from page 26 to page 119.^[33] Anthropic plans to tell investors that advanced AI could pose "catastrophic or existential risks to humanity," and that its models could exhibit "self-preserving behaviors," including attempts to "resist shutdown," to "conceal or manipulate information" and behavior "resembling blackmail."^[17] The filing states: "Our development of highly advanced models, platforms, and applications and expansion of use cases could further increase the risk that our models cause harm."^[17]

Other disclosures reported by Reuters include:

• Evaluation limits. Reuters quoted the prospectus as saying "Potential model awareness of our evaluation efforts creates a significant limitation on our ability to assess model safety," adding that models can develop unexpected capabilities in training that may not be found until after deployment.^[17] In the public S-1 the sentence begins "Such model awareness of our evaluation efforts."^[33]
• Safety spending. Anthropic described safety work as "resource-intensive," said returns on it are unclear, and did not disclose how much it spends on safety research.^[17]
• Release cadence. The company said customer usage, and therefore revenue, is driven by new models, and that a "continuous and overlapping cadence" of releases is "inherent to remaining at the frontier of AI development."^[17]

Reuters set the prospectus against Anthropic's own research showing that increasingly autonomous models can behave in unexpected and potentially harmful ways in controlled tests, including sabotaging code, assisting fraud and manipulating information, and against chief executive Dario Amodei's call for the industry to slow the release of new capabilities. It noted that Anthropic released Claude Opus 5.5 the week before, to counter OpenAI's momentum after GPT-6 Astra.^[14][17]

Valuation

Reuters reported that the sale could value Anthropic at more than $2 trillion, more than double the $965 billion valuation of the May Series H.^[14] The Wall Street Journal had earlier put the expected value at about $2 trillion and the raise at up to $100 billion.^[23] Either figure would exceed SpaceX, whose June IPO valued it at $1.77 trillion.^[14] Reuters reported on October 15 that some investors at Anthropic's investor day had discussed a valuation of around $1.5 trillion.^[43] After the public filing, the Wall Street Journal reported a range of $1.5 trillion to $2 trillion, according to Investing.com.^[40]

Comparison with the public S-1

Sources: Reuters and TechCrunch for the middle column; S-1 and CNN for the right column.^{[14][16][17][31][33][38]}

Each 2025 financial figure Reuters reported matches the public filing within rounding. The visible differences are that the public version names the instruments behind the charge, gives no aggregate obligations figure, fills the third board seat elected by Class A and Class F holders, and has a longer risk factors section.^[33]

Structure and governance

Anthropic is a Delaware public benefit corporation (PBC), a form that lets directors balance the financial interests of stockholders against a stated public benefit purpose. In Anthropic's case that purpose is "the responsible development and maintenance of advanced AI for the long-term benefit of humanity."^[11] According to Reuters, the prospectus says the company will keep operating as a PBC after the IPO, and the public S-1 confirms this.^[16][33] Anthropic was co-founded and is led by siblings Dario Amodei, the chief executive, and Daniela Amodei, the president, who also chairs the board.^[16]

The Long-Term Benefit Trust

A distinctive feature is the Long-Term Benefit Trust, announced in September 2023. The Trust was designed as an independent body of five trustees with backgrounds in AI safety, national security, public policy, and social enterprise, and it holds a special class of stock (Class T) that lets it elect and remove a number of Anthropic's board members. That number phases in according to time- and funding-based milestones, and "in any event, the Trust will elect a majority of the board within 4 years."^[11] Under the public S-1, the Trust's Class T directors make up a majority of the board from the completion of the offering.^[33] The initial trustees were Jason Matheny, Kanika Bahl, Neil Buddy Shah, Paul Christiano, and Zach Robinson. Matheny stepped down in December 2023, Christiano in April 2024, and Bahl and Robinson in January 2026. Anthropic's Trust page says Richard Fontaine joined in May 2025, Mariano-Florentino Cuéllar in January 2026, and former Federal Reserve chair Ben Bernanke in July 2026.^[11] On October 21, 2026, Anthropic announced that physicist Christopher Stubbs had joined, naming the other trustees as Shah, Fontaine and Bernanke.^[41] The S-1 likewise lists four trustees (Shah, chair since June 2023; Fontaine; Bernanke; and Stubbs) and says one of the five seats is vacant; Cuéllar is not among them. The S-1 dates Fontaine's service from April 2025 and Bernanke's from June 2026, a month earlier in each case than the Trust page.^[33]

Founder LLC and share classes

Reuters reported on September 28 that the prospectus creates a new "Founder LLC" made up initially of Anthropic's seven co-founders, whom the filing calls "distinctly equipped to be stewards of our mission."^[16] A majority vote of the seven directs a single share of Class F stock carrying 50.1% of the total voting power on key corporate matters, including the election of some directors.^[16] Members named by Reuters include Dario Amodei, Daniela Amodei, chief compute officer Tom Brown, and Chris Olah.^[16] The main features reported by Reuters are:

Source: Reuters, with notes from the public S-1.^[16][33] Reuters wrote that the structure "could effectively diminish the influence of everyday investors," and that some details had first been reported by The Information earlier in September.^[16] The filing also says Anthropic has "chosen not to develop certain commercially attractive offerings, such as image and video generation models, in order to direct our compute toward our research and safety priorities."^[16]

Earlier analysis of the Trust

In June 2026, Fortune reported on a Harvard Law paper by professor Jesse Fried and S.J.D. candidate Idan Reiter, "AI Corporate Governance and Ben & Jerry's Risk," and interviewed Fried about how the Trust would work for public investors.^[12] Fortune reported that the Trust's powers are more limited than those of the OpenAI Foundation over OpenAI: it can nominate only a majority of the board, not the entire board, certain decisions require a board supermajority, and a supermajority of Anthropic's investors can terminate the Trust and remove the directors it appointed.^[12] The public S-1 sets out the post-IPO thresholds for amending the Trust or redeeming its Class T stock (see "Governance as filed" above).^[33] Fried treated that investor backstop as what makes the arrangement workable, saying "If investors really don't like what the guardians are planning to do, I expect the guardians will back off."^[12] The paper draws a cautionary comparison to the independent board of Ben & Jerry's, whose 2021 decision not to renew the license of the brand's Israeli licensee, Fried and Reiter argue, backfired: Unilever overrode the board in 2022 and achieved the opposite of what the directors intended.^[12] The point cuts both ways: the same kill switch that reassures investors also limits how binding the safety mission really is. Fortune's article predates the Reuters reporting on the Founder LLC, which adds a second controlling bloc alongside the Trust.^[12][16]

Comparison to OpenAI

Much of the coverage framed Anthropic's filing as the opening move in an IPO race with OpenAI.^[2][6] OpenAI completed its conversion to a public benefit corporation in October 2025, with its nonprofit, now the OpenAI Foundation, keeping the power to appoint every director.^[12] In late May 2026, CNBC reported that OpenAI was preparing to file its own confidential prospectus and was looking to go public as soon as September.^[5] On June 8, OpenAI said it had "recently submitted a confidential S-1" and had "not decided on timing yet."^[24] In an interview with Fortune reported on September 12, Altman said "right now would be an ill-advised moment to go public," and, asked about 2026, replied "I would say not 2026, yeah."^[25] Reuters reported on September 28 that OpenAI was expected to list by early 2027, according to media reports.^[14] On October 22, TechCrunch, citing CNBC, reported that OpenAI was working on raising up to $30 billion in private funding at a $1.25 trillion pre-money valuation in a round led by Coatue and investors in the UAE, with terms still being negotiated.^[47]

Fortune reported in June that demand for Anthropic shares in the secondary market had been stronger than for OpenAI's, and that reports suggested OpenAI's financials were not as strong as Anthropic's, though it cautioned that a confidential S-1 "is still a maybe."^[6] Dario Amodei framed the rivalry in revenue terms in a Fortune cover story in December 2025, saying "I would argue it's maybe even the most likely world in which our revenue passes theirs a year from now."^[6] OpenAI's last reported private valuation was $852 billion after its March 2026 round.^[2][5] Reuters wrote that analysts expect the first frontier AI lab to go public to set valuation benchmarks for the industry.^[14] Commenting on the public filing, Emarketer analyst Jacob Bourne told CNN that an anticipated wave of fall IPOs had failed to materialize amid concerns including "AI safety risks that have reportedly contributed to rival OpenAI delaying its own IPO until next year."^[38]

Reception

Coverage of the September prospectus reports focused on the gap between revenue and spending. Reuters itself noted that SpaceX's shares had slipped from their $160 first-day level to around $147, that AI and chip stocks had sold off recently, and that Anthropic's sale "will further test whether enthusiasm for the AI trade can withstand greater scrutiny."^[14] On the risk disclosures, Reuters observed that while companies routinely outline product risks, "few, if any, have issued warnings suggesting their technology could cause potential human extinction."^[17] Stocktwits reported skeptical reactions, including from AI critic Ed Zitron, who pointed to $12.6 billion of spending against $4.6 billion of revenue, and from equity analyst Ross Hendricks, who questioned whether the $518 billion of infrastructure commitments were sustainable given their scale.^[30] TechCrunch reported that the Financial Times had also reviewed the filing, and that according to the FT, Anthropic's second-quarter 2026 revenue alone reached $11.5 billion and the company was on track for a second straight quarter of operating profit on an adjusted basis.^[31] The public S-1 later showed second-quarter revenue of $11.555 billion and Adjusted Operating Income of $2.235 billion for that quarter, after an adjusted loss of $617 million in the first quarter.^[33]

Coverage of the public filing on October 26 focused on growth, losses and the scale of the compute commitments. CNBC highlighted the roughly 900% net revenue retention and compared it with 108% for the median public cloud software company as of January, citing Altimeter Capital partner Jamin Ball; it also noted that neocloud stocks such as CoreWeave and Nebius had lost more than 30% of their value in October.^[37] Redpoint Ventures managing director Erica Brescia told CNBC that Anthropic is "an absolutely blockbuster business," while Axios business editor Dan Primack said the filing signaled a shift toward using public markets to pay for compute: "They're looking for scale, they're going to spend an absolute fortune on compute."^[37] CNN described the filing as detailing "surging business growth, along with a vast multitude of warnings about how it could all go wrong," and reported Dan Ives of Yorkville & Ives Co. as saying the growth would likely overshadow concerns about losses.^[38] Investing.com cautioned that the various commitment figures "should not be treated as a single, additive estimate" of infrastructure spending.^[40] CNBC also reported that the ticker ANT is already used on the Canadian Securities Exchange by Anteros Metals, a six-person, pre-revenue mineral rights company.^[37]

A day later, CNBC described investors as split. Bulls pointed to the roughly 1,300% annual sales growth, the 900% net revenue retention and the 85% implied gross margin; bears pointed to the roughly $80 billion first-half net loss, the compute commitments and more than 90 pages of risk factors.^[42] Janus Henderson portfolio manager Denny Fish called the IPO "a generational moment." David Trainer, chief executive of New Constructs, whose firm had called it the "most ridiculous IPO of 2026," described the offering as "a huge Wall Street grift." Elevation Partners co-founder Roger McNamee said "we should view this S-1 as speculative fiction," and Gil Luria of D.A. Davidson said that if the IPO is priced correctly it will likely do "very well."^[42] Before the filing, investor Chamath Palihapitiya had argued that getting public and raising the capital it needed mattered more than maximizing valuation, writing on X on October 14 that "it's more critical for Ant to get public and get the $200B they need than to optimize for valuation," and that his guess at a market-clearing price was about $1 trillion.^[46]

Timeline and caveats

The sequence of events in 2026 was compressed. Anthropic closed the Series H on May 28, announced the confidential draft S-1 on June 1, and filed five confidential amendments between July and October.^[1][4][35] Over the summer the reported target moved from as soon as October to November or later. On September 28 Reuters published the first detailed look at the prospectus, and on October 26 Anthropic filed it publicly.^{[14][19][21][23][32]}

Several cautions apply as of October 28, 2026:

• Not yet effective. The S-1 is a preliminary prospectus. The registration statement has not become effective, and the share count, price range and several aggregate figures are blank.^[32][33]
• Timing and valuation are reported, not filed. The November listing target and the $1.5 trillion to $2 trillion valuation range come from press reports citing unnamed people, not from the filing.^[39][40][43]
• Run-rates are not revenue. The run-rate figures quoted before the filing come from Anthropic's funding announcements and press reports. The S-1 defines Run-Rate Revenue but leaves its October 31, 2026 value blank, and reports recognized revenue of $4.585 billion for 2025 and $16.286 billion for the first half of 2026.^[5][10][33]
• Commitment figures are not additive. The S-1 discloses compute commitments by agreement and by accounting category at different dates and does not give a combined total, so the figures should not be summed. Totals such as $450 billion, $475 billion or $750 billion come from press reports and calculations, not from the public filing.^{[33][40][44][45]}

Until Anthropic prices the offering and its shares begin trading, the IPO remains a planned event rather than a finished one.

References

1. ^1 ^2 ^3Anthropic, "Anthropic confidentially submits draft S-1 to the SEC," June 1, 2026. anthropic.com/...confidential-draft-s1-sec
2. ^1 ^2 ^3 ^4 ^5 ^6 ^7CNBC, "Anthropic confidentially files IPO prospectus with SEC, prepping Wall Street for landmark AI deal," June 1, 2026. cnbc.com/...anthropic-ipo-s1-prospectus
3. ^1 ^2NPR, "AI giant Anthropic prepares to sell stock to the public; files preliminary IPO paperwork," June 1, 2026. npr.org/...anthropic-ipo-filing-ai-large
4. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11Anthropic, "Anthropic raises $65B in Series H funding at $965B post-money valuation," May 28, 2026. anthropic.com/...series-h
5. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8CNBC, "Anthropic tops OpenAI as most valuable AI startup, nears $1 trillion valuation in latest round," May 28, 2026. cnbc.com/...anthropic-open-ai-startup-value
6. ^1 ^2 ^3 ^4 ^5Fortune, "Anthropic confidentially files its S-1 first, but the IPO race with OpenAI is just beginning," June 1, 2026. fortune.com/...anthropic-s1-confidential
7. ^1 ^2NBC News, "Anthropic secures $965 billion valuation after raising $65 billion," May 28, 2026. nbcnews.com/...ation-raising-65-billion-rcna347390
8. ^TechCrunch, "Anthropic raises $65 billion, nears $1T valuation ahead of IPO," May 28, 2026. techcrunch.com/...-nears-1t-valuation-ahead-of-ipo
9. ^1 ^2Anthropic, "Anthropic raises $30 billion in Series G funding at $380 billion post-money valuation," February 12, 2026. anthropic.com/...-380-billion-post-money-valuation
10. ^1 ^2 ^3 ^4 ^5TechCrunch, "Anthropic files to go public," June 1, 2026. techcrunch.com/...anthropic-files-to-go-public
11. ^1 ^2 ^3Anthropic, "The Long-Term Benefit Trust," September 19, 2023 (updated with later membership changes). anthropic.com/...the-long-term-benefit-trust
12. ^1 ^2 ^3 ^4 ^5 ^6Fortune, "Harvard Law: Anthropic is about to sell a safety mission that Wall Street can veto," June 1, 2026. fortune.com/...dian-ben-jerrys-ice-cream-anthropic
13. ^CNN Business, "Anthropic files to go public in a potentially trillion-dollar debut," June 1, 2026. cnn.com/...anthropic-ipo-filing
14. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13 ^14 ^15 ^16 ^17 ^18 ^19 ^20 ^21 ^22Reuters (via Yahoo Finance), "Exclusive-Anthropic's IPO prospectus shows sweeping AI vision, surging costs," September 28, 2026. finance.yahoo.com/...po-prospectus-shows-231722972
15. ^1 ^2 ^3 ^4 ^5 ^6CNBC, "Anthropic's IPO prospectus shows sweeping AI vision, surging costs: Reuters," September 28, 2026. cnbc.com/...eeping-ai-vision-surging-costs-reuters
16. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12CNBC, "Anthropic leaders to control AI lab to promote public good over market forces: Reuters," September 28, 2026. cnbc.com/...public-good-over-market-forces-reuters
17. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11Reuters (via Yahoo Finance), "Exclusive-Anthropic warns AI may pose 'existential risks to humanity' in IPO filing," September 28, 2026. finance.yahoo.com/...hropic-warns-ai-may-001709847
18. ^1 ^2 ^3Reuters (via Yahoo Finance), "Anthropic's path from AI startup to industry-defining IPO," September 28, 2026. finance.yahoo.com/...ai-startup-industry-235243135
19. ^1 ^2 ^3Investing.com (via Yahoo Finance), "Anthropic said to pick Morgan Stanley and Goldman to lead IPO," June 3, 2026. finance.yahoo.com/...pick-morgan-stanley-175802678
20. ^1 ^2CNBC, "Anthropic moves closer to mega-IPO as bankers line up investor meetings," July 15, 2026. cnbc.com/...anthropic-ipo-banks-investor-meetings
21. ^1 ^2 ^3 ^4CNBC, "Anthropic IPO launch shifts toward mid-October: Reuters," September 5, 2026. cnbc.com/...unch-shifts-toward-mid-october-reuters
22. ^1 ^2 ^3Bloomberg (via Yahoo Finance), "Anthropic Said to Choose Nasdaq for Much-Anticipated IPO Listing," September 13, 2026. finance.yahoo.com/...-choose-nasdaq-much-180359465
23. ^1 ^2 ^3 ^4PYMNTS, "Anthropic Targets November IPO as Revenue Surges," September 18, 2026. pymnts.com/...-targets-november-ipo-revenue-surges
24. ^1 ^2OpenAI, "Confidential submission of draft S-1 to the SEC," June 8, 2026. openai.com/...openai-submits-confidential-s-1
25. ^1 ^2TechCrunch, "OpenAI's Sam Altman says it would be 'ill-advised' to go public in 2026," September 12, 2026. techcrunch.com/...ill-advised-to-go-public-in-2026
26. ^U.S. Securities and Exchange Commission, "SEC's Division of Corporation Finance Expands Popular JOBS Act Benefit to All Companies," June 29, 2017. sec.gov/...2017-121
27. ^U.S. Securities and Exchange Commission, EDGAR company search results for "anthropic," accessed September 29, 2026. sec.gov/...browse-edgar
28. ^1 ^2 ^3 ^4 ^5 ^6 ^7Capital Brief, "Anthropic's IPO prospectus reveals USD4.6b revenue surge, and USD8.1b operating loss," September 29, 2026. capitalbrief.com/...cd-1025-43ee-ab7b-cf4c551fbce5
29. ^1 ^2Ynet News, "Anthropic lost nearly $42 billion in 2025 as it prepares for possible $2 trillion IPO," September 29, 2026. ynetnews.com/...bjsox0u9fl
30. ^Stocktwits (via Yahoo Finance), "Anthropic IPO Prospectus Leaked: $42B Loss, Over Half A Trillion In Cloud Commitments, Skeptics Lash Out," September 29, 2026. finance.yahoo.com/...ospectus-leaked-42b-023245734
31. ^1 ^2TechCrunch, "Anthropic's prospectus details losses, growth, and, yes, a warning that its AI could end humanity," September 28, 2026. techcrunch.com/...g-that-its-ai-could-end-humanity
32. ^1 ^2 ^3 ^4 ^5 ^6 ^7Anthropic, "Anthropic files S-1 with the SEC for proposed initial public offering," October 26, 2026. anthropic.com/...s1-sec
33. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13 ^14 ^15 ^16 ^17 ^18 ^19 ^20 ^21 ^22 ^23 ^24 ^25 ^26 ^27 ^28 ^29 ^30 ^31 ^32 ^33 ^34 ^35 ^36 ^37 ^38 ^39 ^40 ^41 ^42 ^43 ^44 ^45 ^46 ^47 ^48 ^49 ^50 ^51 ^52 ^53 ^54 ^55 ^56 ^57 ^58 ^59 ^60 ^61 ^62 ^63 ^64 ^65 ^66 ^67 ^68 ^69 ^70 ^71 ^72 ^73 ^74 ^75 ^76 ^77 ^78 ^79 ^80 ^81 ^82 ^83 ^84 ^85 ^86 ^87 ^88Anthropic, PBC, Form S-1 Registration Statement (preliminary prospectus), filed with the U.S. Securities and Exchange Commission, October 26, 2026. sec.gov/...anthropicpbc-sx1
34. ^1 ^2Anthropic, PBC, Form S-1 filing index and filing-fee exhibit, accession no. 0001628280-26-067168, October 26, 2026. sec.gov/...0001628280-26-067168-index
35. ^1 ^2 ^3 ^4 ^5 ^6 ^7U.S. Securities and Exchange Commission, EDGAR filing list for Anthropic, PBC (CIK 0002130357), accessed October 28, 2026. sec.gov/...browse-edgar
36. ^Anthropic (@AnthropicAI), post on X announcing the S-1 filing, October 26, 2026. x.com/...2114705042817110293
37. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10CNBC, "Anthropic's IPO plans: 900% revenue retention, but $80 billion in first-half losses" (live updates), October 26, 2026. cnbc.com/...anthropic-ipo-live-updates-ai-s-1-filing
38. ^1 ^2 ^3 ^4 ^5CNN Business, "Anthropic set to make stock market debut as questions swirl about safety of AI," October 26, 2026. cnn.com/...anthropic-files-ipo
39. ^1 ^2Quartz, "Anthropic files to go public, setting up a landmark moment for the AI boom," October 26, 2026. qz.com/...ic-ipo-filing-nasdaq-ant-november-102626
40. ^1 ^2 ^3 ^4 ^5 ^6Investing.com, "Anthropic files for IPO, with revenue surging and potential $2 trillion valuation," October 26, 2026. investing.com/...enue-jumps-to-163-billion-4972842
41. ^1 ^2 ^3Anthropic, "Physicist Christopher Stubbs joins Anthropic's Long-Term Benefit Trust," October 21, 2026. anthropic.com/...bbs-joins-long-term-benefit-trust
42. ^1 ^2CNBC, "Anthropic's IPO could be 'speculative fiction' or a 'generational moment' - investors are split," October 27, 2026. cnbc.com/...hropic-ipo-investors-growth-losses-s-1
43. ^1 ^2 ^3 ^4Reuters (via The Star), "Investors temper Anthropic IPO valuation expectations over revenue growth slowdown, sources say," October 15, 2026. thestar.com.my/...enue-growth-slowdown-sources-say
44. ^1 ^2 ^3 ^4Bloomberg (via Yahoo Finance), "Anthropic May Spend $750 Billion on Computing Costs, Report Says," October 20, 2026. uk.finance.yahoo.com/...pend-750-billion-200421561
45. ^1 ^2Fortune, "Anthropic is on the hook for billions in compute. Its customers can walk away at any time," October 26, 2026. fortune.com/...ts-customers-can-walk-at-any-time-2
46. ^Chamath Palihapitiya (@chamath), post on X, October 14, 2026. x.com/...2110446110540288152
47. ^TechCrunch, "OpenAI working on new $30B raise at $1.25T valuation ahead of IPO," October 22, 2026. techcrunch.com/...-at-1-25t-valuation-ahead-of-ipo