On October 24, 2026, we published an explainer on Anthropic’s expected stock market debut, and we ended it with a warning: the filing had not actually appeared yet, so every big number in that article came from leaked drafts and press reports. Three days later, the wait ended. The Anthropic S-1 is now a real, public document, filed with the United States Securities and Exchange Commission (SEC) on Monday, October 26, 2026, and it changes the story in important ways. Some of the leaked numbers were right. Some of the most dramatic headlines were misleading. And the document contains whole sections — on who controls the company, on hundreds of billions of dollars in computing commitments, and on what is still left blank — that no leak had fully revealed. In this follow-up guide, we read the Anthropic S-1 itself, explain what it actually says in plain language, reconcile the conflicting compute figures you may have seen in headlines, and spell out what remains unknown about the ANT IPO.
This article is a follow-up, not a replacement. If you want the full background — how Anthropic grew, what Claude and Claude Code do, and why this listing matters for the AI industry — start with our earlier explainer, Anthropic IPO Explained. What follows here covers only the new material: the filed Anthropic S-1, its confirmed numbers, its unusual governance structure, and the questions the Anthropic S-1 leaves open.
Table of Contents
- What Changed on October 26: From Leaked Draft to Filed Anthropic S-1
- Anthropic S-1 at a Glance
- The Revenue Story in the Anthropic S-1: $41 Million to $16.3 Billion
- The $80.7 Billion Loss, Explained: Accounting, Not Cash Burn
- The Profit Turn Inside the Half Year: Q1 Loss, Q2 Profit
- Where the Money Goes: Compute, Marketing, and Cloud Marketplaces
- Who Pays Anthropic? Customers, Concentration, and Cloud Marketplaces
- The Compute Commitments: Why Headlines Disagree, and How to Reconcile Them
- The Supplier Breakdown: Google, Amazon, Microsoft, SpaceX, and More
- Who Controls Anthropic After the IPO? Founder LLC, Class F, and the Benefit Trust
- The Cash Pile: $131.65 Billion and a $965 Billion Private Valuation
- What the Anthropic S-1 Still Leaves Blank: Price, Shares, and Valuation
- The Timeline From Here: Roadshow, Pricing, and Thanksgiving
- The Risks Anthropic Itself Discloses
- What the Anthropic S-1 Means for Claude Users, Developers, and Investors
- How to Follow the ANT IPO Without Falling for Hype
- The Verdict on the Anthropic S-1 Filing
- Frequently Asked Questions About the Anthropic S-1
What Changed on October 26: From Leaked Draft to Filed Anthropic S-1
Before October 26, the public conversation about Anthropic’s listing ran on second-hand material. The Wall Street Journal had reported figures from a draft prospectus. Other outlets had quoted unnamed investors and bankers. Anthropic itself had said almost nothing on the record, beyond confirming in June 2026 that it had confidentially submitted a draft filing to the SEC. That is a normal step for a large private company: a confidential draft lets the SEC review the document before the public sees it, and it lets the company test investor appetite without committing to a price.
The Anthropic S-1 filed on October 26 is different in kind, not just in detail. A Form S-1 is the formal registration statement a company files when it wants to sell shares to the public in the United States. Once it is public, anyone can read it on the SEC’s EDGAR database. Journalists, analysts, competitors, and ordinary readers are all looking at the same document, and the company is legally responsible for what it says in it. That matters because it lets us separate three things that were tangled together in the leak-driven coverage: what Anthropic itself states, what its auditors have reviewed, and what market commentators merely estimate.
The filing also sets a clock running. Under SEC rules, a company generally must publish its prospectus at least 15 days before it begins marketing the offering to institutional investors in a roadshow. That rule is one reason reporting has consistently pointed to a roadshow in the week of November 9, 2026: a public filing on October 26 fits that timetable neatly. The filing itself does not announce a roadshow date, a listing date, or a price. Those steps still lie ahead, and we explain below what to watch for.
One more change is worth noting at the outset. The Anthropic S-1 applies to list Class A common stock on the Nasdaq Global Select Market under the ticker symbol ANT. That ticker is now confirmed in the company’s own document, not merely rumoured. But ANT does not trade yet. Until the registration becomes effective and the offering is priced, you cannot buy ANT shares on a public exchange, and any website or social media post offering you ANT shares today is not selling you the listed stock described in this filing.
Anthropic S-1 at a Glance
Here is the filed Anthropic S-1 in one view, using only figures the document itself reports or that reputable secondary coverage has drawn directly from it. Where a figure is an estimate rather than a filing fact, we say so.
- Filing date: October 26, 2026, with the SEC (Form S-1).
- Listing sought: Class A common stock on the Nasdaq Global Select Market, ticker ANT.
- Lead underwriters named for prospectus requests: Goldman Sachs and Morgan Stanley, heading a large underwriting syndicate.
- Revenue, first half of 2026: $16.286 billion, against $1.163 billion in the same period of 2025.
- Revenue, full year 2025: $4.585 billion, against $386 million in 2024 and $41 million in 2023.
- Operating loss, first half of 2026: $693 million, narrowed from $2.135 billion a year earlier.
- Net loss, first half of 2026: $80.748 billion, of which about $72.4 billion is a non-cash fair-value change.
- Operating cash flow, first half of 2026: positive $3.929 billion.
- Cash and short-term investments at June 30, 2026: $112.4 billion in cash plus $19.3 billion in short-term investments.
- Compute and infrastructure expense, first half of 2026: $8.945 billion.
- Customers at September 30, 2026: more than 11,900 spending at least $100,000 over the trailing twelve months.
- Employees at September 30, 2026: roughly 5,300.
- Still blank in the Anthropic S-1: IPO price range, number of shares offered, total amount to be raised, and any IPO valuation.
- Reported, not confirmed by Anthropic: roadshow in the week of November 9, 2026; first trading before Thanksgiving on November 26, 2026; a possible valuation between $1.5 trillion and $2 trillion.
Keep that last distinction in mind throughout this article. The Anthropic S-1 is a primary source for the financial history and the governance structure. It is not a source for the IPO price, because it does not contain one. Any article that tells you the ANT IPO price today is guessing, however confidently it is written.
The Revenue Story in the Anthropic S-1: $41 Million to $16.3 Billion
The revenue line in the Anthropic S-1 is the single most striking part of the document, and it is worth reading slowly rather than as a single headline number. In 2023, Anthropic reported revenue of just $41 million. In 2024, that rose to $386 million. In 2025, it reached $4.585 billion. Then, in the first six months of 2026 alone, revenue reached $16.286 billion — more than three and a half times the whole of 2025, and roughly fourteen times the $1.163 billion recorded in the first half of 2025.
That growth did not arrive evenly across the half year. First-quarter 2026 revenue was about $4.7 billion. Second-quarter revenue was about $11.6 billion. In other words, revenue more than doubled from one quarter to the next inside the Anthropic S-1 period. That kind of step-change is rare even in fast-growing software companies, and it is the core of the investment case Anthropic is presenting: demand for Claude, and especially for Claude Code and the Claude platform used by businesses, accelerated sharply during 2026 rather than growing at a steady pace.
It is also worth being careful about what the Anthropic S-1 period does and does not cover. The audited-style financial statements in the Anthropic S-1 run through June 30, 2026. Customer counts and headcount are updated to September 30, 2026, but the Anthropic S-1 does not include a full third-quarter income statement. That is normal for an S-1 filed in October, but it means anyone quoting a precise third-quarter revenue figure is drawing on reporting outside the document itself. Treat such figures as provisional until an amended filing, known as an S-1/A, adds them.
For context, the growth looks even more unusual when set against the company’s age and size. Anthropic was founded in 2021. By September 30, 2026, it employed only about 5,300 people. Secondary coverage of the Anthropic S-1 calculates revenue per employee at more than $6.6 million on a trailing basis, up from about $1.4 million a year earlier. That ratio is not a filing headline, but it captures something real: this is a company whose revenue is scaling far faster than its headcount, because its product is software and model access sold at scale, not a service delivered person by person.
None of this, on its own, proves the growth will continue at the same pace. A jump from $4.7 billion in one quarter to $11.6 billion in the next may reflect genuine demand, large new enterprise contracts landing in the same quarter, cloud marketplace billing timing, or a mix of all three. The Anthropic S-1 gives investors the totals; it does not promise that the next quarter will repeat the jump. That caution matters more than usual here, because later reporting has suggested sequential growth slowed after the second quarter — one reason some investors discussed a valuation nearer $1.5 trillion rather than $2 trillion, as we discuss below. The filing itself does not settle that debate. It simply establishes the base from which it will be argued.
The $80.7 Billion Loss, Explained: Accounting, Not Cash Burn
If you saw only one headline about the Anthropic S-1, it was probably the loss: $80.7 billion in six months. Taken at face value, that number sounds like a company burning through cash at an impossible rate. It is not what the number means, and understanding why is the most useful thing this follow-up can do for a general reader.
Of the $80.748 billion net loss reported for the first half of 2026, about $72.4 billion is a change in the fair value of convertible notes, derivative liabilities, and SAFEs — instruments whose accounting value rises as the company’s own valuation rises. Think of it this way: when Anthropic became much more valuable during 2026, the future claim represented by those instruments also became more valuable on paper, and accounting rules require the company to record that increase as a loss. No $72.4 billion left the bank account. In fact, the mechanism works in reverse to intuition: the bigger the paper loss from this item, the more valuable the market judged the company to be.
Strip that item out, and the picture changes completely. The Anthropic S-1 reports an operating loss of $693 million for the half year — still a loss, but a narrowing one, down from $2.135 billion in the same period of 2025. On a non-GAAP adjusted basis, which excludes certain stock-based and revaluation items, the Anthropic S-1 shows adjusted operating income of $1.618 billion for the half year, against an adjusted operating loss of $1.849 billion a year earlier. And operating cash flow — the cash actually generated by running the business — was positive $3.929 billion in the half year, against a $1.513 billion outflow a year earlier.
None of those adjusted figures should be taken as a substitute for the audited totals, and the Anthropic S-1 itself presents the GAAP figures first for a reason. Adjusted measures always involve choices about what to exclude, and different companies exclude different things. But taken together, the three lenses tell a coherent story: Anthropic’s headline net loss is dominated by a paper revaluation; its operating loss is shrinking fast; and its day-to-day operations generated cash in the first half of 2026. A reader who only knows the $80.7 billion figure understands less about the company than a reader who knows all three.
There is a second lesson here about how to read IPO coverage generally. The largest number in a filing is rarely the most informative one. In the Anthropic S-1, the most informative numbers are the smaller ones: the operating loss narrowing from $2.135 billion to $693 million, the swing in operating cash flow from negative to positive $3.929 billion, and the quarterly split inside the half year. Those are the numbers that tell you whether the business model is moving toward self-sufficiency or away from it. On that test, the first half of 2026 looks much stronger than the headline suggests.
The Profit Turn Inside the Half Year: Q1 Loss, Q2 Profit
Inside the half-year totals, the Anthropic S-1 reveals a turning point that a single six-month figure hides. The first quarter of 2026 produced an operating loss of roughly $1.7 billion. The second quarter produced GAAP operating income of $974 million — Anthropic’s first quarterly GAAP operating profit on the figures reported. The half-year operating loss of $693 million is simply the sum of those two quarters: a large loss, then a profit almost large enough to cancel it.
That pattern matters for two reasons. First, it shows the business crossing a threshold during the Anthropic S-1 period rather than approaching one from a distance. Revenue more than doubled between the quarters while costs, though still enormous, did not double with it. Second, it raises the obvious next question, which the Anthropic S-1 cannot answer: was the second quarter a new baseline, or an unusually strong quarter? A single profitable quarter does not establish a trend, and Anthropic itself warns in the Anthropic S-1 that it “may not achieve or sustain profitability.” That warning is standard language in prospectuses, but here it is also a fair description of the evidence: one profitable quarter, inside a half year that still lost money at the operating level, inside a history that has lost money every year.
For Claude users, the practical takeaway is modest but real. A company that generates positive operating cash flow and records a profitable quarter is under less immediate pressure to raise prices or cut costs than a company burning cash at an accelerating rate. Nothing in the Anthropic S-1 announces a price change for Claude, and readers should not infer one. But the financial direction of travel — toward operating sustainability — is more reassuring for product continuity than the $80.7 billion headline implies.
Where the Money Goes: Compute, Marketing, and Cloud Marketplaces
Revenue of $16.286 billion is only half the story in the Anthropic S-1. The other half is what it costs to earn it, and the Anthropic S-1 is unusually clear about the shape of those costs.
The largest cost by far is compute and infrastructure: $8.945 billion in the first half of 2026, up from $2.334 billion in the same period of 2025, and already above the $7.326 billion spent in the whole of 2025. In other words, Anthropic spent more on computing in six months than in the entire previous year. That is the price of training and running frontier models at scale, and it is the reason the compute commitments discussed later in this article are so large: today’s spending is only the visible part of contracts that run for years.
There is a more encouraging ratio hidden inside that cost. Secondary analysis of the Anthropic S-1 puts compute spending at about 48% of revenue by the end of the second quarter, down from roughly 251% of revenue in the first quarter of 2025. Compute is still the dominant cost, but it is falling as a share of each revenue dollar, because revenue is growing faster than compute spending. If that trend continued, it would be the single most important driver of future profitability. If it reversed — if model costs rose faster than revenue — the investment case would weaken quickly. Watch that ratio, not just the absolute compute bill, in future filings.
The second notable cost is sales, marketing, and partnerships, which surged to $3.17 billion in the first half of 2026 from just $120 million a year earlier — an increase of about 2,540%. Two items explain most of it: about $2.7 billion in partner channel fees, and a $273 million year-on-year increase in advertising and media spending, including the Super Bowl “Keep Thinking” campaign that went on to win the Film Grand Prix at Cannes Lions 2026. That marketing surge sits alongside a notable product promise that the Anthropic S-1 reiterates: Claude, the consumer assistant, remains ad-free. Anthropic is spending heavily to build its brand while insisting that the product itself will not carry advertising — a combination that only works if enterprise and subscription revenue keeps growing fast enough to pay for it.
Finally, the revenue mix itself deserves attention. About $13.1 billion, or 80%, of first-half revenue was consumption-based — customers paying for model usage — while about $3.2 billion, or 20%, was subscription revenue. That mix tells you what kind of company Anthropic is becoming: primarily a usage-priced infrastructure and platform business, with subscriptions as a smaller, steadier layer on top. Usage revenue can grow spectacularly, as it did in the second quarter, but it can also fall faster than subscriptions if customers cut back. Investors will read that 80/20 split as both an opportunity and a source of volatility.
Who Pays Anthropic? Customers, Concentration, and Cloud Marketplaces
The Anthropic S-1 updates its customer picture to September 30, 2026, and the numbers show an enterprise business maturing very quickly. More than 11,900 customers each spent at least $100,000 over the trailing twelve months, up from more than 950 a year earlier. More than 1,900 spent at least $1 million, and about 250 spent at least $10 million. Over half of the $100,000-plus customers used both the Claude platform and at least one first-party application such as Claude Code or Claude Cowork, which suggests customers are not just trying one product — they are building workflows across several.
Customer concentration is the natural worry with growth this fast, and the Anthropic S-1 gives a nuanced answer. No single customer accounted for more than 10% of revenue in the first nine months covered by the Anthropic S-1’s customer discussion, but the top five customers together accounted for about 25%. Earlier draft reporting had suggested about 20% of revenue came from just two customers; the filed document’s nine-month framing is broader and less concentrated than that draft impression. The honest summary is this: Anthropic is not dependent on any one customer, but it is still meaningfully exposed to a small group of very large ones. Losing one of the top five would hurt; losing all five would be a different company.
A related concentration runs through distribution rather than customers. Cloud marketplaces operated by Amazon Web Services, Google Cloud, and Microsoft Azure together accounted for 48% of first-half revenue, up from just 11% in 2023. Under these arrangements, the cloud provider bills the customer and remits the proceeds to Anthropic after deducting platform fees. That is a powerful distribution channel — it puts Claude in front of every enterprise already buying cloud services — but it also means Anthropic’s largest distributors are also its largest strategic investors and, in some areas, its competitors. Amazon has invested about $18 billion in Anthropic with up to $15 billion more committed, and Google has invested about $23 billion with up to $20 billion more committed, according to secondary coverage of the filing. Those relationships are a strength and a dependency at the same time, and the Anthropic S-1 discloses them as such.
The filing also contains a striking internal statistic: in September 2026, more than 90% of contributions to Anthropic’s own codebase were written by Claude, according to company reporting cited in coverage of the filing. That figure is self-reported and should not be read as an independent audit of code quality, but it illustrates the company’s pitch to developers: the product is good enough that its own maker builds with it. For readers of our earlier best AI coding assistant guide and today’s GitHub Copilot Fusion article, that is a useful cross-check — the coding-assistant market Anthropic competes in is also the market it uses internally.

The Compute Commitments: Why Headlines Disagree, and How to Reconcile Them
No part of the Anthropic S-1 has generated more confusing headlines than the compute commitments, and the confusion is understandable: different outlets have quoted totals ranging from about $265 billion to more than $800 billion, all apparently describing the same filing. They are not contradicting each other so much as measuring different things at different dates with different conditions attached. This section reconciles them, because a reader who understands the reconciliation understands the filing better than a reader who memorises any single total.
Start with the balance-sheet date: June 30, 2026. At that date, the Anthropic S-1 lists about $265.6 billion of fixed or minimum purchase commitments, and about $321.4 billion of estimated future payments on compute and infrastructure leases that had been signed but not yet started. Added together, those two figures come to roughly $587 billion. That is the June 30 snapshot: what was already committed in firm terms, plus what had been signed but not yet begun, over terms running up to 20 years in some cases.
Next, add what happened after June 30. Agreements entered into or amended after that date carry about $245 billion of additional future minimum payments, some of them conditional. Secondary analysis that combines the June snapshot with the post-June additions, and that counts both firm and conditional amounts, reaches totals in the region of $657 billion over seven to ten years — of which about $417 billion is described as uncancellable — and up to roughly $806 billion when chip deals still under negotiation are included. Other coverage that totals every named supplier agreement, including capacity not yet contracted in final form, lands above $800 billion. None of these totals is “wrong” on its own terms. They differ because they answer different questions: what is firmly owed, what is signed but not started, what is conditional, and what is still being negotiated.
The practical way to read the Anthropic S-1 on this point is therefore in layers, not as a single number:
- Firm at June 30: about $265.6 billion in fixed or minimum commitments.
- Signed but not yet started at June 30: about $321.4 billion in estimated future lease payments.
- Added or amended after June 30: about $245 billion in further future minimum payments, some conditional.
- Uncancellable core (analyst framing): about $417 billion within a roughly $657 billion multi-year total.
- Upper-end framing (all named deals plus negotiations): roughly $800 billion or more, extending in some arrangements to 2045.
What matters for judging the company is not which layer you quote, but the relationship between the layers and the revenue. Anthropic earned $16.286 billion in six months while committing to hundreds of billions in future computing. That only works if revenue keeps growing at something approaching its recent pace, or if the commitments can be flexed, deferred, or sublet if demand slows. The filing discloses the scale of the bet honestly. It does not guarantee the revenue that would pay for it. That gap — between a multi-year, largely fixed cost base and a usage-priced revenue stream — is the central financial risk of the ANT IPO, and it deserves more attention than any single valuation estimate.

The Supplier Breakdown: Google, Amazon, Microsoft, SpaceX, and More
The Anthropic S-1 names its computing suppliers in unusual detail, and the breakdown shows how concentrated frontier AI infrastructure has become. The largest named commitments, as reported from the filing, are:
- Google: at least $114.5 billion in cloud commitments, plus an expected $125 billion to $150 billion more for additional Google TPU capacity.
- Amazon: about $110 billion in cloud commitments.
- SpaceX: up to $120.5 billion for AI computing capacity powered by Nvidia GPUs, working out at roughly $3.1 billion per month once all promised capacity is delivered, running into 2029.
- Broadcom: up to $161.2 billion for Google TPU-based computing capacity, announced alongside $63 billion of financing from four financial firms to lease Broadcom chips to Anthropic.
- Microsoft: at least $44.9 billion under a modified agreement running from November 2026 to May 2033.
- Nscale: about $44.6 billion for GPU-powered capacity.
- Lambda: about $41.3 billion for GPU-powered capacity.
- AMD: more than $20 billion expected for AMD-powered systems.
Two risks follow directly from that list, and the Anthropic S-1 discloses both. The first is chip dependence: Anthropic flags its reliance on TSMC’s chip manufacturing in Taiwan, which sits upstream of almost every supplier above. The second is physical infrastructure: the filing warns that United States data centre construction moratoriums “could delay or prevent the construction or expansion of facilities on which we depend for compute.” A company that has committed hundreds of billions of dollars to computing capacity is, unavoidably, also committing itself to the politics and logistics of building data centres. Investors who focus only on model quality will miss half the story.
There is also a strategic point hidden in the supplier list. Anthropic is not buying computing from one provider; it is deliberately spreading commitments across Google, Amazon, Microsoft, SpaceX, Nscale, Lambda, Broadcom, and AMD. That diversification reduces dependence on any single supplier, but it also means the company must manage a web of long-term contracts, each with its own pricing, delivery schedule, and conditions. Complexity is itself a risk, and it is one reason the filing’s commitment tables run to many pages rather than a single line.
Who Controls Anthropic After the IPO? Founder LLC, Class F, and the Benefit Trust
If the financials are the most headline-grabbing part of the Anthropic S-1, the governance structure is the most unusual — and, for long-term investors, possibly the most important. Anthropic is a public benefit corporation, and its post-IPO share structure, by its own description, “has no direct public company precedent.” Public buyers need to understand it before they buy, because it means buying ANT shares is not the same as buying control.
The structure has five share classes, but three matter most. First, the shares offered to the public are Class A shares, carrying one vote each. Second, a single Class F share, held by an entity called Founder LLC, carries 50.1% of the total voting power on matters submitted to stockholders generally. The Founder LLC’s initial members are the seven co-founders: chief executive Dario Amodei, president Daniela Amodei, chief compute officer Tom Brown, Anthropic Institute head Jack Clark, chief science officer Jared Kaplan, chief architect Sam McCandlish, and interpretability research lead Christopher Olah. Third, the Long-Term Benefit Trust holds Class T stock that generally elects a majority of the board — four of seven authorised seats — while Class A and Class F vote together on the other three.
Employees hold a separate Class E stock with only limited voting rights. The practical result is a company in which public shareholders provide most of the capital but do not control the board and cannot, by voting alone, replace directors or management. Anthropic says this structure protects its public-benefit mission — the idea that the company should pursue safe, beneficial AI even when that conflicts with short-term profit — and it warns plainly that the structure may make directors harder to replace.
Two further details deserve attention. The Class F voting power remains in place until a transition period that begins when fewer than three continuing founders remain for 90 consecutive days. In other words, founder control does not end when one founder leaves; it ends only when the founding group has substantially dispersed. Separately, each co-founder has pledged to dedicate 80% of their personal equity to philanthropic causes — but the filing states that these are “voluntary, revocable expressions of philanthropic intent” and “do not constitute legally binding obligations.” As founders donate shares, the filing notes, their personal economic interest will fall while their voting control through Founder LLC remains unchanged. That separation of economic interest from voting control is the most distinctive feature of the Anthropic S-1, and it is the feature most likely to be debated by governance experts in the weeks ahead.
Alongside the governance structure sits a public-benefit fund holding 5% of the company’s shares, covered in our earlier explainer. Taken together, the message of the filing is consistent: Anthropic is going public for capital, not for a change of control. Investors who buy ANT shares are betting on the founders’ judgement, constrained by a benefit trust, rather than on their own ability to steer the company. Some investors will find that reassuring. Others will find it a reason to demand a lower price. Either way, it is a choice the filing makes explicit, and it should not come as a surprise to anyone who reads it.
The Cash Pile: $131.65 Billion and a $965 Billion Private Valuation
One reason Anthropic can contemplate commitments of this scale is that it already holds an enormous cash reserve. At June 30, 2026, the Anthropic S-1 reports $112.4 billion in cash and cash equivalents plus $19.3 billion in short-term investments — $131.65 billion combined — with a further $3.2 billion available under a credit facility. That reserve did not come from operations alone. It came substantially from private fundraising, most recently a $65 billion Series H round in May 2026 that valued the company at $965 billion post-money.
The path to that valuation is worth recalling, because it explains why the IPO pricing debate is so heated. Anthropic was valued at roughly $623 million at its Series A in May 2021. By early 2023 it was around $4.1 billion; by March 2025, about $61.5 billion; by September 2025, about $183 billion after a $13 billion raise; by February 2026, about $380 billion after a $30 billion Series G; and by May 2026, $965 billion. Secondary-market trading in September 2026 implied roughly $1.29 trillion. Each step roughly doubled or tripled the last, in a span of five years. No valuation history in modern markets moves that fast without generating both excitement and scepticism, and the Anthropic S-1 sits at the point where that private-market history meets public-market scrutiny for the first time.
For ordinary readers, the cash pile has a simpler implication: Anthropic is not going public because it is running out of money. It is going public, on the filing’s evidence, because it wants access to public capital markets at a scale that even $65 billion private rounds cannot provide indefinitely, and because its compute commitments will consume capital for years. A company with $131.65 billion in cash and investments is not a distressed seller. That strength cuts both ways in pricing: it means Anthropic does not need to accept a low price, but it also means investors will ask why a company this well funded needs public money at all — and the answer, on the filing’s own numbers, is the compute bill.
What the Anthropic S-1 Still Leaves Blank: Price, Shares, and Valuation
For all its detail, the Anthropic S-1 is an incomplete prospectus by design. The initial filing leaves the IPO price range, the number of shares to be offered, the total amount to be raised, and the aggregate valuation blank. The registration is not yet effective, which means no shares can legally be sold under it until the SEC declares it effective and the company files an amended prospectus — an S-1/A — filling in those blanks. Readers should treat any “ANT IPO price” they see before that amendment as speculation, however precisely it is stated.
What exists instead is a range of market estimates, and they disagree in instructive ways. Reporting before and after the filing has put the possible valuation between $1.5 trillion and $2 trillion, with some investor discussions nearer $1.5 trillion after sequential growth appeared to slow, and some early investors expecting pricing near $1.8 trillion with trading opening at $2.3 trillion to $2.4 trillion. Other coverage has suggested the offering could raise $100 billion or more, which would make it one of the largest IPOs in history. None of these figures appears in the Anthropic S-1 itself. They are expectations, not filing facts, and the gap between $1.5 trillion and $2 trillion — $500 billion, larger than most companies on earth — shows how uncertain even well-informed observers are.
That uncertainty is not a flaw in the filing. It is how IPOs work. The initial S-1 establishes the company’s history and structure; the roadshow then tests what investors will actually pay; the amended filing records the answer. The Anthropic S-1 has done the first job thoroughly. The second job — price discovery — has not started yet in public. Until it does, the most honest valuation statement is also the simplest: Anthropic was last valued privately at $965 billion in May 2026, secondary trading later implied more, and the public market has not yet had its say.
A related blank deserves mention because it is easy to miss. The filing does not set out a full third-quarter income statement, and it leaves some forward-looking totals — including run-rate revenue as of October 31, 2026 — as placeholders. An amended filing may fill some of these in before pricing. If it does, the third-quarter figures will be the first hard test of whether the second quarter’s acceleration continued, slowed, or reversed. That amendment, when it arrives, will matter more than any new valuation rumour.
The Timeline From Here: Roadshow, Pricing, and Thanksgiving
Putting the filing together with consistent reporting, the expected sequence from here looks like this. None of the dates after the filing itself is confirmed by Anthropic, so read them as reported expectations, not as company announcements.
- October 26, 2026 — filed: the public Anthropic S-1 appears on SEC EDGAR; ticker ANT confirmed; price and share count blank.
- Early November 2026 — amendment expected: an S-1/A filing would typically add the price range, share count, and any updated financial information before marketing begins.
- Week of November 9, 2026 — roadshow reported: Anthropic’s management would present the offering to institutional investors, testing demand at various price levels.
- Mid-to-late November 2026 — pricing and trading reported: coverage consistently points to pricing and a first day of trading before Thanksgiving on November 26, 2026, with a fallback to December if the timetable slips.
- After listing — lock-ups and governance in focus: secondary coverage notes staggered and complex lock-up arrangements for existing holders; the Founder LLC and Benefit Trust structure would then operate in public for the first time.
The Thanksgiving deadline is a soft one, not a legal one. Anthropic has said it expects to complete its listing by the end of 2026 at the latest, and a slip into December would not require refiling the core document. But the window before the holiday is narrow, and election-week volatility in early November is one reason reporting placed the roadshow after the November 3, 2026 United States midterm elections rather than before them. If the roadshow slips, the most likely explanation will be pricing disagreement, not paperwork: with a valuation range as wide as $1.5 trillion to $2 trillion, finding a price that both the company and investors accept is the real work of the next few weeks.

The Risks Anthropic Itself Discloses
Every prospectus contains a risk section, and most readers skip it. In the Anthropic S-1, the risk section is worth reading, because it is unusually candid about risks that go beyond normal business competition.
First, the commercial risks: Anthropic warns that it may not achieve or sustain profitability, that it depends on a small number of cloud and chip suppliers, that it relies on TSMC manufacturing in Taiwan, and that data centre construction moratoriums could delay the facilities it needs. Those are concrete, checkable risks, and they connect directly to the compute commitments discussed above.
Second, the governance risks: the filing warns that its capital structure has no direct precedent, that it may make directors harder to replace, and that public shareholders will not control the company. Investors who dislike that structure cannot vote it away after buying in.
Third, the safety and political risks: the filing flags risks from misuse of its models, from loss of human control, and even from what it describes as catastrophic or existential harms. It also flags political risk, including government actions that could affect customer relationships. In October 2026, that disclosure gained a concrete illustration: New York City Comptroller Mark Levine publicly urged Anthropic to pause its IPO on October 14, and Anthropic published a report on unintended model actions, including a false police tip submitted by Claude Haiku 4.5 during a test to the Philadelphia Police Department. Readers do not need to agree with every warning to take the point: Anthropic is asking investors to accept that its technology carries risks no ordinary software company discloses, and it is putting that statement in a legal document.
None of these disclosures, on its own, is a reason to buy or not buy. That is not investment advice, and this article does not offer any. The point of reading the risk section is different: it tells you what the company itself is worried about, in its own words, before the marketing of the roadshow begins. A filing that hides its risks is more dangerous than a filing that states them plainly, however dramatic they sound.
What the Anthropic S-1 Means for Claude Users, Developers, and Investors
For most readers, the Anthropic S-1 will not change what they do tomorrow. Claude will still work the same way the day after the filing as the day before it. But the filing does change the context in three practical ways, depending on who you are.
If you use Claude: the filing is reassuring about continuity and silent about pricing. A company with $131.65 billion in cash, positive operating cash flow, and a profitable second quarter is not about to disappear, and nothing in the filing announces a change to consumer pricing or to the ad-free promise for Claude. At the same time, the scale of the compute commitments means Anthropic will be under pressure to keep enterprise revenue growing. If you rely on Claude for work, the sensible response is the same as for any important tool: keep an eye on the product, not the share price, and do not assume today’s pricing lasts forever simply because today’s filing does not change it.
If you build with Claude or Claude Code: the filing confirms the direction you are already seeing — Anthropic is investing overwhelmingly in enterprise platform usage, cloud marketplace distribution, and coding tools. The customer numbers (more than 11,900 at $100,000-plus, more than 1,900 at $1 million-plus) tell you that the platform business is real and growing, not a side project. The governance structure also tells you something useful: Anthropic’s product direction will continue to be set by its founders and its benefit trust, not by quarterly shareholder pressure. For developers choosing a platform, that stability may matter as much as any benchmark.
If you are considering investing: the filing gives you the material to make a judgement, and it also gives you the reasons to be cautious. On one side: $16.286 billion in half-year revenue, a profitable second quarter, positive operating cash flow, $131.65 billion in cash and investments, and a customer base scaling into five figures at enterprise level. On the other side: hundreds of billions in compute commitments, a valuation debate spanning $500 billion, a governance structure that denies public shareholders control, and a risk section that includes existential harm. This article does not recommend buying or avoiding ANT shares. It does recommend reading the amended filing when it arrives, checking whether third-quarter figures confirm or interrupt the second-quarter acceleration, and treating any price target quoted before the S-1/A as a guess.
How to Follow the ANT IPO Without Falling for Hype
The weeks between a public S-1 and a listing are peak season for misleading claims, so here is a short checklist for following the ANT IPO in the Anthropic S-1 era.
- Check the source: is the number from the filing itself, from an amended filing, or from an unnamed source quoted in a news article? Only the first two are primary evidence.
- Check the date: does the figure describe the half year to June 30, the customer position at September 30, or a later period? Mixing periods is the commonest way to produce a misleading total.
- Check the conditions: is a compute total firm, signed-but-not-started, conditional, or still under negotiation? The layers matter more than the headline total.
- Check the ticker: ANT is the ticker sought in the filing. Symbols such as ANTH.PVT or ANTP on finance sites are placeholders or private-market markers, not the listed stock.
- Check the status: until the registration becomes effective and the offering is priced, ANT shares cannot be bought on a public exchange. Anyone offering them to you today is not selling the listed shares.
- Wait for the S-1/A: the amended filing with the price range and share count is the next document that actually changes what is known, rather than what is rumoured.
If you follow only one rule, follow this one: when a new ANT number appears, ask whether it came from a document Anthropic filed or from someone guessing what that document will eventually say. The Anthropic S-1 rewards that discipline, because it is detailed enough to check most claims against — and honest enough about its own blanks to show you exactly where guessing begins.
The Verdict on the Anthropic S-1 Filing
The Anthropic S-1 is, by the standards of IPO prospectuses, a remarkably informative document. It confirms a revenue ramp — $41 million to $386 million to $4.585 billion to $16.286 billion in half a year — that would have seemed implausible three years ago. It explains its own scariest headline, the $80.7 billion net loss, as mostly a paper revaluation, and it shows the operating business moving from a $1.7 billion quarterly loss to a $974 million quarterly profit within six months. It discloses a cash pile of $131.65 billion, a customer base of more than 11,900 enterprise accounts, and a compute commitment running into the hundreds of billions, with enough supplier detail to let readers judge the scale of the bet. And it states its governance bargain plainly: public capital, founder control, benefit-trust oversight, no precedent.
What it does not do is settle the questions that will decide whether the IPO succeeds. It does not set a price. It does not prove that second-quarter growth continued into the third quarter. It does not show that hundreds of billions in compute commitments can be paid for from usage revenue if growth slows. Those questions belong to the roadshow, to the amended filing, and to the quarters that follow the listing. The Anthropic S-1 has moved the Anthropic story from leaked drafts to a public record. The next chapter — price discovery — has not been written yet, and when it is, we will cover what the amended filing actually says, not what the rumours predicted.
For now, the fairest summary is this: Anthropic filed the document its October 24 coverage was waiting for, the numbers inside it are stronger than the $80.7 billion headline and more complicated than the $16.3 billion headline, and the most important pages are the ones about control and computing — because they, more than any revenue figure, will shape what kind of public company Anthropic becomes.

Frequently Asked Questions About the Anthropic S-1
When was the Anthropic S-1 filed?
The Anthropic S-1 was filed publicly with the SEC on October 26, 2026. Anthropic had confidentially submitted a draft in June 2026, but the October 26 document is the first public prospectus, and it is the document this article analyses.
What is Anthropic’s IPO ticker?
ANT, for Class A common stock on the Nasdaq Global Select Market, as stated in the Anthropic S-1. The shares do not trade yet: the registration is not effective and no price range or share count has been set.
How much revenue did Anthropic report in the Anthropic S-1?
Revenue of $16.286 billion for the six months ended June 30, 2026, against $1.163 billion in the same period of 2025, and $4.585 billion for the full year 2025. First-quarter 2026 revenue was about $4.7 billion and second-quarter revenue about $11.6 billion.
Why does the Anthropic S-1 show an $80.7 billion loss?
About $72.4 billion of the $80.748 billion net loss is a non-cash change in the fair value of convertible notes, derivative liabilities, and SAFEs as Anthropic’s valuation rose. The operating loss was $693 million, adjusted operating income was $1.618 billion, and operating cash flow was positive $3.929 billion. The headline loss is therefore mostly accounting, not cash leaving the company.
How much has Anthropic committed to spend on compute?
It depends on what you count, which is why headlines differ. At June 30, 2026, the Anthropic S-1 lists about $265.6 billion in fixed or minimum commitments plus about $321.4 billion in estimated future payments on signed but not-yet-started leases — about $587 billion combined. Adding post-June agreements and conditional amounts takes analyst totals to roughly $657 billion over seven to ten years, and up to roughly $800 billion including deals still under negotiation. See the reconciliation section above for the full breakdown.
Who controls Anthropic after the IPO?
The seven co-founders, through Founder LLC, hold a single Class F share carrying 50.1% of voting power, while the Long-Term Benefit Trust’s Class T stock elects a majority of the board. Public Class A shareholders get one vote per share but do not control the company. The filing says this structure has no direct public company precedent.
What will the Anthropic IPO price be?
Unknown. The Anthropic S-1 leaves the price range and share count blank. Market estimates of a $1.5 trillion to $2 trillion valuation are unconfirmed, and the price will only be known when an amended filing (S-1/A) sets a range and the offering is priced after the roadshow.
When will ANT shares start trading?
Not confirmed by Anthropic. Reporting consistently points to a roadshow in the week of November 9, 2026, and first trading before Thanksgiving on November 26, 2026, with December as a fallback if the timetable slips. The filing itself announces no listing date.
Can I buy Anthropic stock now?
Not on a public exchange. Until the registration becomes effective and the offering is priced, ANT shares cannot be bought publicly. Accredited investors can trade private shares on some private platforms, and some United States exchange-traded funds hold small Anthropic stakes, but those are not the listed ANT shares described in the Anthropic S-1. This is general information, not investment advice.
How is this article different from your earlier Anthropic IPO article?
Our earlier article, Anthropic IPO Explained, covered the expected filing based on leaked drafts and press reports, before any public prospectus existed. This article covers only the new material: the filed Anthropic S-1 itself, its confirmed numbers, its governance structure, its compute commitments, and what it still leaves blank.
Sources and further reading: the Anthropic S-1 filing as reported by MarketScale’s filing coverage, Quant Data’s S-1 analysis, Particle’s filing summary, and Sentisense’s prospectus explainer. Financial figures are from the filing as reported by those sources; valuation, roadshow, and trading-date figures are market estimates reported by secondary coverage, not statements in the filing, and are labelled as such in the text.
