🔍 Read the full analysis: The Risks Anthropic Flags As It Prepares To Go Public on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Anthropic filed a preliminary IPO registration statement on October 26, 2026, disclosing rapid revenue growth alongside substantial compute commitments and governance limits for public shareholders. The filing reports an $80.7 billion net loss for the first half of 2026, most of it tied to non-cash changes in the value of financing instruments, and warns that its AI products could pose catastrophic or existential risks.
Anthropic, PBC filed a preliminary registration statement for an initial public offering with the U.S. Securities and Exchange Commission on October 26, disclosing fast-growing revenue, large future computing obligations and a governance structure that limits public shareholders’ influence. The filing also warns that the company’s AI products could “pose catastrophic or existential risks to humanity,” making the prospectus a rare public account of both the commercial and safety risks facing a frontier AI developer.
The filing reports $16.3 billion in revenue for the six months ended June 30, 2026, compared with $1.16 billion in the same period a year earlier. Second-quarter revenue was $11.56 billion. Anthropic recorded its first quarterly operating profit in Q2, at $974 million, but still reported a net loss of $80.7 billion for the first half.
That loss is not equivalent to $80.7 billion in cash spending. Anthropic attributes $72.4 billion primarily to the accounting increase in the fair value of convertible notes, derivatives and other financing instruments as the company’s valuation rose. First-half operating cash flow was positive at $3.9 billion, and the company reported $131.7 billion in cash and short-term investments as of June 30.
Anthropic’s filing describes extensive commitments to computing capacity, cloud services and data-center infrastructure. A widely reported $542 billion total does not appear as a stated figure in the preliminary prospectus: the relevant aggregate field is blank, and figures used to construct the headline refer to different categories and periods. The company says non-cancelable purchase commitments rose from $54.6 billion at the end of 2025 to about $265.6 billion at June 30, with additional obligations signed after that date.
The filing also sets out a structure in which a trust elects a majority of the board, while a founder share carries 50.1% of the votes. Public investors would receive one vote per share within the remaining voting power. The preliminary document does not disclose the offering size or price; reported valuation targets and descriptions of the deal as a record-setting IPO are not confirmed terms in the filing.
Anthropic's IPO filing, the numbers
From $376 million to $11.6 billion a quarter
Six quarters of revenue, the shrinking compute share, and why a company with a first operating profit reports an $80.7 billion loss.
Revenue per quarter
USD, as filed. Q3 2026 is left blank in the preliminary prospectus.
Compute and infrastructure, % of revenue
Revenue grew faster than the committed compute bill.
Anthropic expects margins to “compress” again when large blocks of new compute come online (p. 134).
Where the $80.7 billion loss comes from
First half of 2026. Operating cash flow was positive: +$3.9B.
Source: Anthropic, PBC, Form S-1, filed 26 Oct 2026 (pp. 11, 23, 25, 134, 158–159, 169; Note 8, p. F-40). Gross margin: implied from Note 8, same calculation as CNBC. Not investment advice.
Powered by Thorsten Meyer AI
The Cost of Scaling Anthropic
The filing gives customers, investors and business partners a more detailed view of the risks attached to relying on a fast-growing AI supplier. Anthropic’s sales are largely tied to usage: $13.1 billion of first-half revenue came from consumption-based services, while $3.2 billion came from subscriptions. That model can expand quickly when demand rises, but it also connects revenue to continuing customer use and to the infrastructure needed to serve it.
That infrastructure creates a financial exposure that may not move in step with sales. Anthropic says computing capacity is bought 12 to 18 months ahead and warns that margins may compress as large capacity blocks become available. For companies using Claude in products or operations, the commitments and the stated strategy of making products difficult to leave are relevant to questions about supplier dependence and continuity. The filing does not establish that service will be disrupted; it does make the scale and timing of the company’s capacity needs more visible.
For prospective shareholders, rapid growth does not by itself settle whether current spending can produce durable profits. The reported Q2 operating profit is a meaningful change, but the six-month operating result remained a loss, and the company’s adjusted margin excludes several costs. The filing’s governance provisions also mean that buying shares would not necessarily give investors voting power proportionate to their economic stake.
The risk disclosure about potential catastrophic or existential harm matters beyond company finances. It places the possible consequences of AI products within the company’s own investor-facing account of material risks. The prospectus does not say such harm has occurred or predict that it will; it identifies the possibility as a risk investors should weigh.
Products Worth Considering
Samsung 27-Inch Odyssey QD-OLED G8 4K 240Hz Gaming Monitor
Canon SELPHY CP1500 Wireless Compact Photo Printer with RP-108 Ink and Paper Set
Canon imagePROGRAF PRO-1100 Professional 17-Inch Wireless Inkjet Photo Printer Bundle
As an affiliate, we earn on qualifying purchases.
What the Filing Actually Reports
Anthropic’s reported revenue rose from $41 million in 2023 to $386 million in 2024 and $4.585 billion in 2025. The first-half 2026 total of $16.286 billion marks a steep acceleration, though the filing leaves third-quarter revenue blank. The figures describe different reporting periods and should not be treated as a complete full-year 2026 result.
The company’s loss requires similar care. Its first-half net result includes large fair-value changes related to financing instruments, rather than only ordinary operating expenses. Anthropic reported a first-half operating loss of $693 million, while Q2 alone produced an operating profit. Its own adjusted operating margin for the half was 10%, a non-GAAP measure that excludes stock-based compensation, non-cash charitable contributions and a $1.5 billion copyright settlement. These measures are not interchangeable.
Reports about a $542 billion compute figure combine separate disclosures, including future minimum payments under agreements made or amended after December 31, 2025, and an estimate for additional TPU capacity in 2028–2031 whose terms are not final. The filing also records about $245 billion in further obligations signed after June 30, plus about $21 billion in short-term contracts. Because the categories are defined differently and may overlap, they should not be added together as a single confirmed bill.
IPO Terms and Compute Totals
The October 26 document is preliminary. Offering size, share price and final valuation are not disclosed, and the filing does not confirm reported estimates of a $1.5 trillion to $2 trillion valuation. It is not yet clear when the IPO will be priced or how many shares will be offered.
The full scale of future computing costs also remains difficult to state as one figure. The prospectus leaves its aggregate committed-spend amount blank, while reported totals combine agreements and estimates with different definitions. Some planned capacity arrangements, including the estimate for additional TPU capacity in 2028–2031, do not yet have final terms. The filing’s warnings identify risks, but do not establish how likely or severe any particular outcome will be.
Watch for Amended IPO Disclosures
Investors and customers can expect further detail if Anthropic files amended registration materials, including any completed offering terms and updated financial information. Until then, the share price, offering size and final valuation remain unknown. Subsequent disclosures may also clarify capacity agreements and the blank aggregate commitment figure.
As the IPO process develops, readers should distinguish reported figures from the company’s own disclosures, and accounting losses from cash flows. The next filings and any pricing announcement will show how Anthropic presents the risks to potential investors and whether the proposed governance and voting arrangements remain as described.
Key Questions
Has Anthropic set an IPO price or offering size?
No. The preliminary registration statement does not state an offering size or share price. Reported valuation estimates are not confirmed terms in the filing.
Why did Anthropic report an $80.7 billion first-half loss?
Anthropic said $72.4 billion was primarily a fair-value change in convertible notes, derivatives and other financing instruments as its valuation increased. The company also reported positive operating cash flow of $3.9 billion for the period, so the net loss should not be read as an equivalent cash outflow.
Does the filing confirm a $542 billion compute bill?
No. That figure is a reported sum, not a total printed in the preliminary filing. The prospectus lists different commitment figures and estimates, some of which overlap or are not final, and leaves the aggregate committed-spend figure blank.
Will public shareholders control Anthropic?
The filing describes a trust that elects a majority of the board and a founder share with 50.1% of the votes. Public shareholders would receive one vote per share within the remaining voting power, so their influence would be limited under the structure described.
What AI safety risk does Anthropic disclose?
The prospectus says the company’s products could “pose catastrophic or existential risks to humanity.” This is a risk disclosure about potential harm, not a statement that such harm has occurred.
Source: ThorstenMeyerAI.com
