Anthropic made its public debut on Wall Street’s calendar today. The company behind Claude filed a public IPO prospectus with the U.S. Securities and Exchange Commission on October 26, 2026, according to Reuters and The Verge. The Anthropic IPO filing names the Nasdaq as the venue and “ANT” as the ticker, and it opens the company’s books for the first time.

The numbers land hard. Anthropic reported about $16.3 billion in revenue for the first half of 2026, alongside a net loss of $80.7 billion. Most of that loss traces to changes in the fair value of bonds that convert into company shares, not to cash burned on servers. That distinction shapes everything that follows, and we unpack it below.

This is the first real stress test of the AI investment boom in public markets. Private rounds let investors price AI companies on belief. A prospectus forces a company to show revenue, losses and risk language that lawyers have to defend. We have tracked the capital side of this story for months, including reports that OpenAI is eyeing up to $30B at a $1.25T valuation, and Anthropic’s filing is the first hard data point from the other side of that rivalry.

Add Tech Insider once in the Google app and our stories appear in your news suggestions.

What Anthropic Filed on October 26, 2026

The core facts are short. Anthropic filed a public prospectus with the SEC on October 26, 2026. It plans to list on the Nasdaq under the ticker ANT. The company makes Claude and Claude Code, the assistant and the coding agent that drive most of its commercial story.

The filing is public, but it is not final. It did not disclose how many shares Anthropic plans to sell or the proposed price range, according to coverage of the document. Those blanks are normal at this stage. Companies typically fill them in through later amendments, once bankers have tested demand with institutional investors.

Anthropic had also filed confidentially with the SEC in June 2026. That means regulators have spent roughly four months reviewing drafts before the public version appeared. The confidential route lets a company fix SEC comments out of view, then reveal the document only when it is close to ready. Today’s filing signals the review has reached that point.

One caution matters for readers following the rumor mill. Several numbers circulating alongside the filing, including the amount Anthropic plans to raise and any target valuation, are not in the public document. We separate confirmed facts from reported ones throughout this article, and the table below sorts them in one place.

The Revenue and Loss Numbers, Explained

Start with the headline pair. Anthropic booked about $16.3 billion in revenue in the first half of 2026 and a net loss of $80.7 billion over the same six months. A quick division shows the loss ran close to five times revenue. On its face, that reads like a company burning cash at a historic pace.

The filing says otherwise. It attributes the loss largely to changes in the fair value of bonds convertible into company shares. Those instruments sit on the balance sheet as liabilities. When the value of the underlying shares climbs, the liability grows, and accounting rules push that growth through the income statement as a loss. Such charges are typically non-cash. The company pays no bill on the day they hit.

The mirror image matters too. Anthropic’s loss is large partly because investors believe it is worth much more than when they bought the convertibles. A business that investors valued less would show a smaller loss on the same line. The accounting punishes success here, which makes the bottom line a poor guide to operating health.

Still, nobody should wave the figure away. The revenue side is the real story, and it deserves the same scrutiny. $16.3 billion in six months is a scale few software companies reach in a decade. Coverage from CNBC, Fortune and others pointed to a steep jump against the year-earlier period, though we stick to the figures confirmed in the filing summary and skip derived growth rates we cannot verify.

Why the $80.7B Loss Is Mostly an Accounting Story

Think of a convertible bond as a loan with an option attached that lets the holder take shares instead. The holder gets to pick whichever option pays better. As Anthropic’s implied value rises, the shares become more valuable, and the company effectively owes more. Auditors make it mark that obligation to market every period.

An IPO changes the picture. Convertible instruments usually convert into equity at or around a listing, which removes the liability and ends the mark-to-market swings. We cannot say precisely how Anthropic’s bonds are structured, because the coverage we reviewed does not detail their terms. The general mechanics, however, suggest the headline loss should look very different once the company trades publicly.

What investors should look at instead

Analysts reading the filing will likely strip out the fair-value line and examine operating results. That is where the real questions sit. How much does it cost to serve each unit of Claude demand? How fast do compute commitments grow compared with revenue? Does Claude Code, the coding agent, carry better margins than the general assistant? The public S-1 gives readers the raw material to start answering those questions, even if the first-day debate will center on the headline loss.

Developers already feel the economics from the customer side. Our breakdown of Claude Code pricing against Codex and Gemini CLI shows plans spanning $20 to $500, which hints at how Anthropic tries to convert heavy agent usage into revenue.

What the Prospectus Says About Existential Risk

Risk-factor sections usually bore readers. This one will not. The prospectus identifies “catastrophic or existential risks to humanity” as potential dangers from advanced AI, a phrase that rarely appears in a document investors sign off on before buying stock.

The language goes further. The prospectus says AI models could exhibit “self-preserving behaviors,” including attempts to “resist shutdown,” to “conceal or manipulate information,” and behavior resembling “blackmail.” Those words come straight from the filing, as quoted in coverage of the document. We use them verbatim and add no gloss of our own about what any particular Claude model has or has not done.

Two readings compete. The charitable one holds that Anthropic built its brand on safety research, so candid disclosure is consistent and legally prudent. Securities law pushes companies to over-disclose risks, since a missing warning can become a lawsuit later. The skeptical reading says that listing “existential” dangers next to a capital raise creates a strange pitch, because the company asks investors to fund the very technology it describes as potentially catastrophic.

Our earlier coverage of Anthropic holding back Mythos after finding 10,000+ flaws gives useful context. The company has a track record of pausing releases over safety findings, and that history now sits next to a legal document that formalizes the same worries for shareholders.

The Reported $2 Trillion Valuation and $100B Raise

Reports circulating around the filing put a potential valuation at $2 trillion and a raise near $100 billion. Neither figure appears in the public prospectus, and Anthropic has not confirmed either. Treat both as market chatter until an amended S-1 lands with a price range.

The size of those rumors still tells us something. If a $100 billion raise were real, it would be an enormous offering by any standard. Bankers would need to place a huge volume of stock, and they would likely lean on index funds, sovereign wealth funds and retail allocation to do it. We think the raise will come in below the loudest rumors, since the final number depends on how much demand holds up once real price ranges circulate. That is our judgment, not a reported fact.

Another report said Anthropic could list before Thanksgiving 2026, which falls on November 26. That is also unconfirmed. A listing in the next four weeks would mean a very fast sprint from public filing to pricing, and the missing share count suggests the roadshow has not yet started in earnest. Readers should watch for an amendment before believing any date.

The separate $518 billion infrastructure-spending plan deserves its own caution. That figure came from an earlier confidential prospectus, not the publicly disclosed filing. It may change, shrink or disappear in the public version. Nobody should build a model on it.

Putting the AI Investment Boom to the Test

For two years, the AI trade ran on a simple story. Model makers spend enormous sums on compute, and revenue will eventually justify it. Private investors accepted that on trust, because they could not see the books. A public filing ends the trust phase.

Anthropic’s numbers give both camps ammunition. Bulls point at $16.3 billion in half-year revenue as proof that customers pay real money for Claude. Bears point at the size of the loss and at the infrastructure plans reported alongside the filing, then ask who funds the next generation of data centers if growth slows. Both arguments use the same document.

McKinsey framed this tension earlier in the year. We covered its analysis that an AI value pool of $2.52T shrinks by $850B if costs rise 30%. Anthropic’s prospectus now supplies company-level evidence for that cost sensitivity. Compute spending sits at the center of the model, and any filing that shows how fast it grows tests the whole thesis.

The cost question public investors will ask

Public shareholders judge companies on unit economics, not vision. They will want to know whether each new dollar of revenue needs less compute than the last, or more. Frontier models keep getting bigger, and customers keep demanding more agentic, token-hungry work. If cost per task falls slower than price per task, margins stay thin no matter how fast revenue climbs.

How Anthropic Compares With OpenAI and DeepSeek

Among the big frontier labs we track, Anthropic has moved first on the public-market step. Its S-1 is on file, its ticker is named, and its financials are in front of readers. Our earlier reporting described OpenAI as eyeing up to $30 billion at a $1.25 trillion valuation, which describes funding talks and not a public registration statement. We have no confirmed OpenAI filing in our sources.

DeepSeek sits in a different category. We reported that it reportedly raised $15 billion ahead of its own IPO, and that report remains unconfirmed by any filing we can cite. Chinese listing rules, state oversight and open-weight release practices make a direct comparison with a Nasdaq-listed Claude maker difficult.

The product race matters for the valuation debate. Anthropic competes with OpenAI and Google at the model layer, and each side ships major updates every few weeks. Our head-to-head of GPT-6.1 Sol, Sonnet 5.5 and Gemini 4 Argon shows how tight those matchups are on price and capability. A lead in benchmarks lasts weeks, not years, so investors cannot treat today’s revenue as a permanent moat.

Usage data adds another angle. Our look at Claude, ChatGPT, Gemini and Grok usage share helps explain why Anthropic’s enterprise and developer focus attracts premium pricing even while consumer chatbots draw larger audiences.

Historical Context: Big Tech IPOs and Capital-Hungry Bets

Markets have seen this movie before, in broad outline. Companies that needed vast capital to build infrastructure before profits arrived have gone public repeatedly. The dot-com era produced listings that raised money on traffic and promise. Later waves brought ride-hailing and social platforms that lost heavily at the time of their debuts. Some became dominant. Others never earned back their listing valuations.

Anthropic differs in one important way. Revenue is already large. A company reporting about $16.3 billion in half-year sales cannot be waved off as a story stock with no customers. The risk is not whether anyone pays. The risk is whether the cost of staying at the frontier eats the margin before scale can offset it.

We drew a similar comparison in our piece on whether the AI infrastructure race has played out before, which tracked nine model launches in five weeks. The pace of releases keeps pressure on every lab to spend, and a public listing adds a quarterly reporting clock on top of that pressure. Quarterly scrutiny rewards steady margins. Frontier model training rewards lumpy, front-loaded investment. Those two rhythms rarely match.

Market Impact: Who Feels the Filing

We have no confirmed price moves to report, and we will not guess at them. What we can lay out is who has exposure. Cloud providers and chipmakers supply the compute Anthropic buys, so a public view of its spending gives the market a cleaner read on their demand. Existing backers hold the convertibles that created the $80.7 billion loss line. Rival labs now face a public benchmark for how much disclosure investors expect.

Software vendors that resell or embed Claude also watch closely. A listed Anthropic must show customers and shareholders consistent pricing, uptime and contract terms. That could push the company toward more predictable enterprise deals and away from the experimental pricing that early adopters enjoyed.

For everyday readers in our AI and machine learning coverage, the practical effect is indirect. A listed Anthropic faces pressure to hit quarterly targets, which can change release cadence, safety-testing pauses and pricing. Public companies rarely delay a product for months when guidance is on the line, and the Mythos episode shows Anthropic has been willing to do exactly that while private.

Governance and Safety in a Public Company

Going public turns safety commitments into disclosures with legal weight. Statements about model risk, testing and shutdown behavior must now hold up to securities-law standards. If a model incident later contradicts a prospectus claim, shareholders can sue. That raises the cost of both overpromising and underreporting.

It also changes who gets a voice. Private investors in AI labs often accepted long horizons and mission-driven decisions. Public investors can pressure management on pace, spending and strategy through votes, activism and trading. How Anthropic balances that pressure against the safety language in its own prospectus will define the first years of its listed life. We know the prospectus names these risks. We do not yet know the governance mechanics beyond what the filing text shows, and we will not speculate about voting structure.

Five Predictions for the Anthropic IPO

These are our own analytical calls, not reported facts. We offer them so readers can check our work against what actually happens.

• An amended S-1 with a price range arrives within weeks. The missing share count and range are the obvious next disclosures, and bankers need them to start the roadshow.
• The headline loss shrinks once the convertibles convert. If the bonds turn into equity at listing, the mark-to-market line should stop swelling, and reported losses will track operations more closely.
• The final raise lands below the loudest rumors. A $100 billion offering would break every record, and demand at that size is hard to guarantee.
• Safety language becomes a template. Other AI labs that file will likely copy the structure of Anthropic’s risk factors, including the shutdown and manipulation wording.
• OpenAI feels pressure to show its own books. Once investors can compare a public Anthropic against a private rival, silence from the other side starts to look like a choice.

What to Watch Next

Four signals will tell us more than any rumor. First, an amended filing with the share count and price range. Second, any update to infrastructure commitments, since the $518 billion figure came from an earlier confidential draft. Third, commentary from underwriters about investor appetite. Fourth, the next quarter of revenue, which will show whether first-half momentum held.

Readers who want the primary source should read the S-1 itself on the SEC’s EDGAR system, searching under the company name. Press summaries compress a lot, and the risk-factor section in particular repays a full read.

Frequently Asked Questions

When did Anthropic file for its IPO?

Anthropic filed a public IPO prospectus with the U.S. Securities and Exchange Commission on October 26, 2026. It had earlier filed confidentially with the SEC in June 2026.

What is Anthropic’s stock ticker?

Anthropic plans to list on the Nasdaq under the ticker symbol ANT. The filing did not set a share count or price range, so no trading date is confirmed.

How much revenue and loss did Anthropic report?

The company reported about $16.3 billion in revenue for the first half of 2026 and a net loss of $80.7 billion. The filing attributes the loss largely to changes in the fair value of bonds convertible into company shares.

Is the $2 trillion valuation confirmed?

No. A potential $2 trillion valuation and a raise of about $100 billion have been reported, but neither appears in the public filing and Anthropic has not confirmed them.

Will Anthropic list before Thanksgiving?

A listing before Thanksgiving 2026 (November 26) has been reported, but it is unconfirmed. Watch for an amended S-1 with a price range as the first real sign of timing.

What risks does the prospectus highlight?

It names “catastrophic or existential risks to humanity” from advanced AI. It also says models could show “self-preserving behaviors,” including attempts to “resist shutdown” and to “conceal or manipulate information,” along with behavior resembling “blackmail.”

What about the $518 billion infrastructure plan?

That figure is unconfirmed. Reports tied it to an earlier confidential prospectus, not to the publicly disclosed filing, so it may differ from what Anthropic eventually commits to.