Anthropic Files Public S-1, Plans to List on Nasdaq as ANT

Anthropic has moved from the shadows of a confidential IPO filing into public view. On October 26, 2026, the company announced on its own newsroom that it had filed a registration statement on Form S-1 with the U.S. Securities and Exchange Commission for a proposed initial public offering, and SEC filing records confirm the document landed on EDGAR that same day. The public S-1 discloses, for the first time in an official filing, Anthropic's intent to list Class A common stock on the Nasdaq Global Select Market under the ticker symbol "ANT," with Goldman Sachs and Morgan Stanley named on the prospectus cover as the banks fielding requests for copies of the offering document.

The filing is a procedural step up from the confidential draft S-1 that Anthropic first submitted to the SEC months earlier, a move pandromeda.com previously reported on when the company's 2025 loss and revenue figures first leaked. Going public with the S-1 means the document — and the financial detail, risk disclosures and governance structure inside it — is now a matter of public record rather than a draft shared privately with regulators.

From a confidential draft to a public filing

Anthropic's path to this point began with a confidential submission to the SEC, a legal mechanism that lets a company have its registration statement reviewed by regulators before any of its contents become public. That earlier draft, filed in the first half of 2026, followed a $65 billion Series H funding round that valued the company at $965 billion post-money — a figure pandromeda.com covered in detail in a comparison of Anthropic's and OpenAI's valuation and IPO trajectories. Confidential status let Anthropic negotiate with the SEC over disclosure questions without tipping its hand to competitors, customers or the public about its finances.

That changed on October 26. Filing an S-1 in public form is the step that starts the public clock on an IPO: it opens the company's books, risk factors and corporate structure to scrutiny from investors, journalists and rivals alike, and it is generally followed by amendments, SEC comment letters, and eventually a roadshow before shares actually begin trading. Anthropic's own announcement used the standard cautionary language that accompanies every S-1: the registration statement "has not yet become effective," meaning no shares can legally be sold, and no offer to buy can be accepted, until the SEC completes its review.

What the public filing discloses that the draft didn't

The headline change from the confidential draft to the public S-1 is simply that the document exists where anyone can read it. Anthropic's newsroom post and the EDGAR filing itself confirm three details that were previously unconfirmed or rumored: the specific exchange (Nasdaq), the proposed ticker ("ANT"), and the class of stock being offered to the public (Class A common stock, implying at least one other class exists with different rights). None of those were nailed down in Anthropic's earlier, confidential-stage public statements, which spoke only in general terms about "a proposed initial public offering."

CNBC's coverage of the filing, published the same day under the headline "Anthropic's IPO plans: Record losses, AI risks and growth potential outlined in the S-1," notes that the document lays out, in regulatory detail, the scale of Anthropic's losses, the breadth of its AI-specific risk disclosures, and the trajectory of its revenue growth — all now attached to a legally binding filing rather than to leaked drafts or anonymous sourcing.

Underwriters and what still isn't set

Goldman Sachs and Morgan Stanley are named on the S-1's cover as the contacts for investors seeking a copy of the preliminary prospectus, a standard signal that they are among the lead underwriters managing the offering. What the filing does not yet contain is just as notable: Anthropic has not disclosed how many shares it plans to sell or at what price range, meaning there is no implied market valuation attached to the filing itself. Those numbers typically appear in a later amendment, closer to the start of the roadshow, when the underwriting syndicate and the company agree on a price band to show institutional investors.

That gap hasn't stopped speculation. Some reports have pointed to private conversations among investors about valuations as high as $2 trillion, with other accounts suggesting bankers are now weighing something closer to $1.5 trillion given the scale of losses disclosed elsewhere in the filing. Both figures are unconfirmed market chatter, not numbers Anthropic has put in writing, and they should be read as such until a price range actually appears in a filing.

The financial picture now on the record

Reuters, reviewing the filing, reported that Anthropic's net loss for the first half of 2026 reached roughly $80.7 billion, a figure dominated by a roughly $72 billion noncash accounting charge tied to the rising fair value of convertible notes the company has issued — instruments that could eventually convert into Anthropic shares rather than cost the company cash. Stripped of that accounting effect, the operating picture is far less dramatic than the headline loss suggests, though still deeply unprofitable by any conventional measure. Revenue for the same six-month period came in at roughly $16.3 billion, continuing the explosive growth pandromeda.com detailed when the smaller, full-year 2025 figures first surfaced from the confidential draft.

Those numbers mark a significant jump from what was previously known. The confidential-stage reporting pointed to a roughly $42 billion net loss and far smaller revenue for all of 2025; the new, half-year 2026 figures in the public S-1 show both metrics scaling up sharply in the months since, a reflection of how quickly Anthropic's enterprise and API business — and its compute costs — have grown.

A third of the filing is risk warnings

Reuters' review of the document also found that risk factors take up roughly a third of the roughly 260-page filing — well above what's typical even for a large tech IPO — including warnings about the possibility that AI systems could pose catastrophic or existential risks, that models could behave in unexpected or deceptive ways, and that heavy reliance on a small number of large customers and cloud partners creates concentration risk. Those disclosures are legally required boilerplate in the sense that every IPO prospectus must list material risks to the business, but the sheer volume and the specific language about AI safety make Anthropic's filing unusual among recent technology listings.

Founder voting control

The S-1 establishes more than one class of common stock, with Class A shares being the ones offered to public investors. Reuters and other outlets, citing reporting from The Information over the summer, have described plans for a separate class of stock carrying additional voting power intended to keep a combined, controlling voting stake with Anthropic co-founder and CEO Dario Amodei and his fellow co-founders, even though Amodei's own economic ownership stake is reported to be in the low single digits. The exact voting mechanics, including any conditions under which that arrangement could lapse, were not independently confirmed by pandromeda.com in the portions of the filing reviewed for this story, and should be treated as reported rather than fully verified pending direct confirmation of the relevant sections of the prospectus.

Compute commitments behind the losses

Much of Anthropic's spending — and much of the reason its losses remain so large even as revenue grows — traces back to compute. Reuters has reported that the company has committed to spending at least $518 billion over roughly a decade on cloud computing and infrastructure, with a large share of that committed spending non-cancelable regardless of how much capacity Anthropic actually uses. That scale of forward commitment is itself a risk factor the filing calls out, since it locks in enormous fixed costs years before the revenue to cover them necessarily materializes.

A public benefit corporation going public

One structural detail confirmed on the SEC filing index is easy to miss but matters for how Anthropic will be governed once it trades: the company is incorporated in Delaware as a public benefit corporation, not an ordinary C-corporation. Under Delaware law, a public benefit corporation's directors are required to balance shareholders' financial interests against a stated public benefit purpose when making decisions, rather than treating shareholder value maximization as the sole duty. Very few companies of Anthropic's scale have taken that structure into public markets, and the arrangement means that, even after shares of Class A stock begin trading on Nasdaq, Anthropic's board will remain legally obligated to weigh factors beyond share price — including, the company has said in the past, its stated mission around AI safety — when it makes major decisions.

That legal structure sits alongside the separate, multi-class voting arrangement reported around the filing. Together, the two mechanisms — a PBC charter plus concentrated founder voting power — are designed to give Anthropic's leadership and board more insulation from short-term shareholder pressure than a typical newly public technology company would have, at least on paper. How that plays out once Anthropic is answering to public markets, analysts and quarterly earnings calls is something that can only be tested after the company actually lists.

Why the public filing matters beyond Anthropic

Anthropic's move to a public S-1 is also being read as a marker for the wider AI industry. Coverage of the filing has repeatedly framed it as a race against OpenAI, Anthropic's largest direct competitor in frontier AI models, to be the first of the major generative-AI labs to actually complete a U.S. listing rather than simply announce intent to do so. Pandromeda.com's side-by-side look at how Anthropic's and OpenAI's valuations, revenue and IPO plans compare lays out how closely matched the two companies' paths to public markets have become over the course of 2026, with both now disclosing eye-popping compute costs and losses alongside eye-popping revenue growth.

For enterprise customers, competitors and regulators, a public S-1 also means Anthropic's business is now subject to a level of ongoing disclosure obligation — quarterly and annual reports, material-event filings, auditor sign-off — that a private company developing frontier AI systems simply does not face. That shift in transparency, as much as any single number in the filing, is arguably the more durable consequence of October 26's announcement.

What happens next

Filing a public S-1 does not set a listing date. Anthropic's registration statement still has to clear SEC review, which typically involves one or more rounds of comment letters and amended filings before the agency declares the statement effective. Only after that can the company and its underwriters launch the investor roadshow where the actual share count and price range are set, usually followed within one to two weeks by the stock's first day of trading. Multiple outlets have reported that Anthropic had been weighing an October listing before shifting its sights toward November 2026, partly so it can show prospective investors a full third quarter of financial results alongside the figures now in the S-1; Anthropic itself has not confirmed a specific listing date. For readers tracking the broader timeline, pandromeda.com's earlier overview of what's known so far about Anthropic's IPO date and valuation lays out how the schedule has shifted over the course of 2026.

What is now fixed, regardless of when trading actually begins, is that Anthropic's financials, risk disclosures and governance plans are public documents rather than leaked drafts. Any further changes — to the price range, the share count, or the risk language — will now show up as amendments to this same public filing on EDGAR, rather than as a second confidential submission.

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