Anthropic vs OpenAI: Valuation, Revenue and IPO Plans Compared

Anthropic's most recent official valuation is $965 billion, set in a $65 billion Series H round the company announced on May 28, 2026. OpenAI's most recent official valuation is $852 billion, set in a $122 billion round it announced on March 31, 2026 — and as of this writing, OpenAI has not announced anything more recent. Both companies have since confidentially filed paperwork with the SEC for a possible IPO: Anthropic on June 1, 2026, OpenAI a week later on June 8, 2026. Neither filing is public, and neither company has confirmed the detailed revenue and loss figures that have circulated in press reports since. This piece separates what each company has actually said, in its own words, from what outlets including Reuters and the Wall Street Journal have reported from documents the companies have not released.

The headline numbers

Anthropic said on May 28, 2026 that it had raised $65 billion in Series H funding, valuing the company at $965 billion post-money. 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. It included $15 billion of previously committed hyperscaler investment, among it $5 billion from Amazon. In the same announcement, Anthropic said its run-rate revenue had "crossed $47 billion" earlier that month — a run-rate figure, which annualizes a recent period of revenue rather than reporting trailing full-year revenue, and is not directly comparable to the audited 2025 revenue later reported from its IPO paperwork.

OpenAI's most recent official figure predates that by two months. On March 31, 2026, OpenAI announced it had closed a funding round of $122 billion in committed capital, lifting its post-money valuation to $852 billion — up from the $730 billion level reported ahead of that round and from $110 billion in initially flagged commitments. SoftBank co-led, with participation reported from Andreessen Horowitz, D. E. Shaw Ventures, MGX, TPG, T. Rowe Price Associates, Amazon, and Nvidia. Financial press covering the announcement, including Reuters and Bloomberg, reported that OpenAI cited roughly $2 billion in monthly revenue and about $13.1 billion in revenue for 2025 — figures attributed to the company by that coverage rather than independently confirmed in a document Pandromeda could verify directly on OpenAI's own site. OpenAI has not announced a newer valuation since.

That gap matters for how this comparison should be read: Anthropic's $965 billion figure and OpenAI's $852 billion figure are not simultaneous snapshots. Anthropic's number is two months newer. Measured strictly by each company's own most recent disclosure, Anthropic is the more highly valued of the two — but neither company has published a value as of October 2026, and both figures are already months old in a market where AI valuations have moved quickly.

Anthropic's Series H, in its own words

Anthropic's announcement framed the $65 billion raise around three uses: safety and interpretability research, expanding compute capacity, and scaling products and partnerships. It did not publish a balance sheet, income statement, or detailed loss figures alongside the funding news — only the run-rate revenue claim. The $965 billion figure followed a $380 billion valuation reported after a $30 billion round that closed in February 2026, led by GIC and Coatue; Anthropic's own newsroom does not appear to carry a dedicated post for that February round, so the $380 billion figure here is as reported by financial press rather than independently confirmed on anthropic.com.

Three days after the Series H announcement closed out May, Anthropic moved toward a public listing. On June 1, 2026, the company said in its own announcement: "Anthropic, PBC confidentially submitted a draft registration statement on Form S-1 to the U.S. Securities and Exchange Commission for a proposed initial public offering of our common stock." The post states plainly that the filing "gives us the option to go public after the SEC completes its review," that "the proposed initial public offering will depend on market conditions and other factors," and that "the number of shares to be offered and the price have not yet been set." No revenue, profit, or valuation figures appear anywhere in that filing announcement.

The financial detail that has circulated since — about $4.6 billion in 2025 revenue, up roughly twelvefold from about $386 million in 2024, against a net loss near $42 billion, most of it a non-cash charge tied to convertible financing instruments, with an operating loss closer to $8 billion — comes from Reuters' review of the draft S-1 itself, not from Anthropic. Pandromeda covered that filing in detail separately, including the $518 billion in reported compute commitments and the roughly 25% revenue concentration in two customers. Anthropic declined to comment on those specifics when Reuters put them to the company, and no public S-1 has yet appeared on SEC EDGAR, meaning the underlying document remains unverifiable against a primary filing as of this writing. More recent press reports, citing a later draft, put Anthropic's first-half 2026 net loss even higher — but those too rest on reporting about an unreleased document, not a company statement, and should be read as unconfirmed.

OpenAI's $852 billion round — still the latest official figure

OpenAI's March 2026 round was, by dollar amount, the largest it has disclosed. The company's own announcement confirmed the $122 billion committed-capital figure and the resulting $852 billion post-money valuation. SoftBank's co-lead role is confirmed across OpenAI's statement and subsequent coverage; Amazon's and Nvidia's roughly $50 billion and $30 billion commitments, and the detail that a large slice of Amazon's investment is contingent on OpenAI reaching an IPO or a technical milestone, come from financial press coverage of the close rather than from a line Pandromeda could directly verify in OpenAI's own release text.

Eight days after Anthropic's confidential filing, OpenAI followed with its own. OpenAI's post, dated June 8, 2026, says the company is disclosing the filing itself because it expected the news to leak otherwise, and — in language that closely mirrors Anthropic's — states that no timing has been decided and that the filing "gives it the option to go public sooner if that proves best." As with Anthropic's notice, OpenAI's filing announcement carries no revenue, loss, or valuation figures; it is a procedural disclosure made under Rule 135 of the Securities Act, not a financial update.

Unlike Anthropic, OpenAI has not had an equivalent Reuters-reviewed prospectus leak as of this writing that Pandromeda could independently corroborate. Reports citing leaked financial documents and shareholder materials — describing roughly $13 billion in full-year 2025 revenue, a 2025 net loss in the tens of billions tied largely to the accounting effects of OpenAI's 2025 conversion from nonprofit control to a public-benefit-corporation structure, and roughly $5.7 billion in first-quarter 2026 revenue — have circulated, but they trace to unofficial leaks and secondary aggregation rather than a single corroborated review comparable to Reuters' Anthropic reporting. OpenAI has not confirmed these figures, and this article treats them as unconfirmed rather than repeating them as fact.

Separately, OpenAI chief executive Sam Altman has publicly played down a near-term listing. In a reported interview, he said a public offering would be "ill-advised" given safety and alignment work underway, and specifically said he would not expect a listing "in 2026" when asked. That is a public statement from Altman as reported by the outlet that conducted the interview, not a company press release, but it is consistent with OpenAI's filing notice: a confidential S-1 preserves the option to list without committing to a date.

Funding history, side by side

The table below lines up what each company has officially disclosed, plus the figures each has had attributed to it in financial press coverage of the same announcements. Where a number is not a company's own statement, it is marked as reported.

Revenue and losses: what each company has actually disclosed

It is worth separating three categories of financial claim about each company, because they get blended together in a lot of coverage: (1) figures a company states in its own announcement, (2) figures from a regulatory filing that outlets have reviewed and reported on, even though the filing itself isn't public, and (3) figures from unofficial leaks or aggregator summation with no clear sourcing chain back to a filing or company statement.

For Anthropic, category one is thin: the only revenue figure in the company's own words is the "$47 billion run-rate" line from the Series H announcement. Category two is substantial: Reuters' review of the draft S-1 is the basis for the $4.6 billion 2025 revenue and roughly $42 billion net loss figures, and that review also produced detail on operating expenses ($12.65 billion), compute spending ($7.33 billion), cash on hand ($20.28 billion as of December 31, 2025), and customer concentration (about 25% of revenue from two customers). Those numbers carry real weight because Reuters describes reviewing an actual draft document, even though Anthropic has not released it and declined to comment on the specifics.

For OpenAI, category one (the company's own words) includes only the valuation and round-size figures from its March and June announcements — no revenue or loss figures appear in the text of either post as Pandromeda could verify it. The widely cited $13.1 billion 2025 revenue and $2 billion monthly run-rate figures sit in category two at best: they were reported by outlets covering the March round and attributed to the company, but without the kind of single, named-reporter document review that underpins the Anthropic loss figures. Later figures — a reported ~$38.5 billion 2025 net loss, a ~$20.9 billion operating loss, and roughly $5.7 billion in first-quarter 2026 revenue — sit closer to category three: they trace to leaked shareholder materials and aggregator reporting that this article could not independently corroborate against a single verifiable primary account, so they are omitted from the comparison table above and should be treated with more skepticism than the Anthropic figures.

The practical upshot: Anthropic's loss and revenue picture for 2025 is better sourced, even though unofficial, than OpenAI's equivalent numbers for the same period. That is a statement about sourcing quality, not necessarily about which company's underlying business is larger or healthier.

Two confidential IPO filings, eight days apart

Both companies' moves toward a listing followed the same legal mechanism and arrived within a little over a week of each other. Anthropic's notice came first, on June 1, 2026; OpenAI's followed on June 8, 2026, with OpenAI's own post noting it was disclosing the filing pre-emptively because it expected news of it to surface anyway. Both announcements are framed under Rule 135 of the Securities Act of 1933 — a provision that lets a company announce a securities offering without that announcement itself counting as an offer to sell. Both explicitly decline to give a timeline, a share count, or a price.

From there, the public record thins out and press reporting takes over. For Anthropic, Reuters has reported — citing people familiar with a later draft — that a listing could come after the US midterm elections in November 2026, with a potential valuation target reported anywhere from roughly $1.5 trillion to above $2 trillion depending on the report and date. For OpenAI, reported underwriter involvement (Goldman Sachs and Morgan Stanley have both been named in press accounts) and a possible listing window between late 2026 and 2027 have circulated, while Altman's own comments point toward 2027 at the earliest. None of this is confirmed by either company, and both companies' own public statements stop at "the filing gives us the option" — neither has stated an actual listing date.

What's confirmed vs. what's reported: a quick reference

Confirmed, in each company's own words: Anthropic's $965 billion valuation and $65 billion Series H size; Anthropic's "$47 billion" run-rate revenue claim; Anthropic's confidential S-1 filing on June 1, 2026; OpenAI's $852 billion valuation and $122 billion round size; OpenAI's confidential S-1 filing on June 8, 2026. Reported, via financial press covering filings or reviewing documents neither company has released: Anthropic's ~$4.6 billion 2025 revenue and ~$42 billion net loss (Reuters' review of a draft S-1); OpenAI's ~$13.1 billion 2025 revenue and ~$2 billion monthly run-rate (press coverage of the March announcement); both companies' IPO timing and target valuations (press reporting, in some cases attributed only to unnamed people familiar with the matter). Readers should weight the first category far more heavily than the second, and the second more heavily than any single-outlet leak that other outlets have not corroborated.

Why this comparison is messier than it looks

Three things complicate a clean side-by-side. First, timing: Anthropic's valuation is current as of late May 2026 and OpenAI's as of late March 2026 — a two-month gap during which AI valuations moved substantially, so the $965 billion-vs-$852 billion comparison is really "Anthropic two months ago vs. OpenAI five months ago," not two simultaneous numbers. Second, what counts as "revenue": Anthropic's own disclosed figure is an annualized run-rate ($47 billion), which projects a recent short period forward and will differ from audited trailing revenue reported elsewhere (the ~$4.6 billion full-year-2025 figure, which reflects actual 2025 performance, not a projection from mid-2026). Comparing a run-rate to a trailing annual figure, or either company's run-rate to the other's trailing figure, will produce misleading conclusions if the distinction isn't kept in view. Third, loss accounting: a large share of both companies' reported net losses, where reported at all, reflects non-cash fair-value adjustments tied to convertible financing and warrant-like instruments rather than operating cash burn — Anthropic's reported operating loss (~$8 billion) is a small fraction of its reported net loss (~$42 billion) for exactly this reason, and similar dynamics have been reported, though less rigorously confirmed, for OpenAI.

For context on how Anthropic has been deploying capital outside of pure model development, Pandromeda has also covered its $150 million commitment to the Genesis Mission science initiative, and on the product side, how Claude Opus 5.5 is priced and positioned gives a sense of where that revenue is coming from on the Anthropic side.

What to watch next

Three concrete things would move this comparison from "reported" to "confirmed." First, either company filing a public S-1 with the SEC — searchable on EDGAR's full-text search — which would convert reported revenue and loss figures into audited, legally attested numbers. Second, either company issuing a new funding or valuation announcement on its own newsroom, the way Anthropic did in May and OpenAI did in March; absent that, any valuation figure attributed to either company after those dates should be treated as a reported estimate, not an update. Third, a stated IPO timeline from either company's own leadership or press office, rather than reporting attributed to unnamed sources. Until one or more of those happens, the most defensible summary is the one this article opened with: Anthropic's last confirmed valuation ($965 billion, May 2026) is higher than OpenAI's last confirmed valuation ($852 billion, March 2026), both companies have confidentially filed IPO paperwork without setting a date, and most of the headline-grabbing loss and revenue figures attached to either company are reported, not confirmed.

Written byTheo Park

Theo Park runs the AI desk at Pandromeda. He follows model launches from the frontier labs and the open-weight community, tracks the assistants and developer tools built on them, and explains what each release changes on pricing, capability and safety. His reporting leans on primary sources: model cards, technical reports, API documentation and the companies' own announcements.

More from Theo ParkAll AI coverage