Anthropic IPO Explained: $17B Revenue, $895B Bet

Anthropic, the company behind Claude, is about to test just how much investors still believe in the AI boom. On November 2, 2026, it filed an amended prospectus for its planned stock market listing, and the numbers inside are startling in both directions: revenue is growing at a speed few companies ever reach, while the bill for the computing power behind that growth is growing even faster. This is the Anthropic IPO in one filing — huge sales, huge losses, and a bet on AI infrastructure worth almost $900 billion.

Quick answer: Anthropic plans to list on the Nasdaq under the ticker ANT, but its November 2 filing sets no share price, no share count, and no IPO date. It reported $16.96 billion in revenue for the three months ended September 30, 2026, an operating loss of $567 million, a net loss of $23.29 billion, and about $895.4 billion in compute and infrastructure commitments. Reports point to a listing before Thanksgiving, but that date is not confirmed by the company.

What happened on November 2?
The numbers that matter
Why is Anthropic losing money while revenue soars?
What is the $895.4 billion commitment?
What valuation are investors talking about?
How does Anthropic compare with OpenAI and Google?
What does the IPO mean for Claude users?
How to read an IPO filing without getting lost
The honest risks
FAQs
Our verdict

What happened on November 2?

Anthropic first filed its public S-1 registration statement with the U.S. Securities and Exchange Commission on October 26, 2026. An S-1 is the document a company files before selling shares to the public for the first time. The November 2 amendment, known as S-1/A, added something investors had been waiting for: a full third quarter of financial results, plus a total for the spending commitments the first filing had left blank.

What it did not add matters just as much. The amended filing still leaves the price range, the number of shares, the amount to be raised, and the listing date blank. In other words, Anthropic has told the market how its business performed, but not yet what it thinks a share is worth. That is normal at this stage — the price range usually arrives in a later amendment, just before the roadshow where the company pitches investors.

Timing is still a matter of reports, not confirmation. Under U.S. rules, a company that filed confidentially must make its S-1 public at least 15 days before its roadshow, which puts the earliest possible roadshow date at November 10. Reuters reported on November 3 that Anthropic is expected to go public before Thanksgiving, and that an Australian retail share offer is planned alongside the U.S. listing, with no separate listing on the Australian Securities Exchange. Treat both dates as reported, not promised: the filing itself gives no date, and market conditions can still change the plan.

The numbers that matter

The headline is the growth. Revenue reached $16.96 billion in the three months ended September 30, 2026, up from $11.56 billion in the previous quarter and $1.38 billion in the same quarter a year earlier. That is roughly twelve times the year-earlier figure, and a 47% jump in a single quarter. For the first nine months of 2026, revenue totalled $33.24 billion. Put the broader AI boom in context with our roundup of 60 key AI statistics for 2026, and Anthropic's curve still stands out — this is not steady growth, it is a near-vertical line.

The second number to watch is what Anthropic spent to earn that revenue. Compute and infrastructure costs hit $9.86 billion in the quarter, or 58% of revenue, up from 48% in the previous quarter. Operating expenses rose to $17.5 billion. The result was an operating loss of $567 million, after an operating profit of $974 million in the second quarter. The swing, the filing says, came primarily from $5.0 billion of short-term compute bought at a premium, plus a $2.0 billion non-cash charge for matching employee donations.

Measure	Q3 2025	Q2 2026	Q3 2026
Revenue	$1.38 billion	$11.56 billion	$16.96 billion
Compute and infrastructure cost	$2.05 billion	$5.56 billion	$9.86 billion
Operating income / (loss)	($3.09 billion)	$0.97 billion	($0.57 billion)
Net loss	($12.14 billion)	($66.31 billion)	($23.29 billion)

Source: Anthropic's amended S-1 filing, as reported by Reuters and in analyses of the filing. Quarterly figures for 2026 are unaudited.

Why is Anthropic losing money while revenue soars?

There are two different losses in this filing, and confusing them is the easiest mistake a reader can make. The operating loss of $567 million is the closer measure of the day-to-day business: revenue minus the cost of running the company, including the compute that powers Claude. The net loss of $23.29 billion is far larger, but it includes big non-cash accounting items. For the nine months ended September 30, Anthropic reported a net loss of $104 billion, including about $94.3 billion in fair-value changes on convertible notes, derivative liabilities, and SAFEs — paper revaluations of financing instruments as the company's value rose, not cash walking out the door.

That does not make the operating picture comfortable. Anthropic itself warns that more short-term compute buying will squeeze margins in the near term. Short-term capacity costs significantly more than capacity bought under multi-year deals, and Anthropic bought a lot of it in the third quarter. The question for investors is simple to state and hard to answer: was that a one-off scramble for capacity, or a sign that demand keeps outrunning the cheaper, long-term supply Anthropic has already secured? The next quarterly update will tell us more than any forecast can.

It is also worth remembering what Anthropic is. Founded in 2021 by former OpenAI researchers led by chief executive Dario Amodei, the company built its name on Claude and on a public commitment to AI safety. That reputation is part of the investment story, and part of the risk: safety-first positioning wins trust with cautious enterprise customers, but it can also mean saying no to uses of the technology that rivals accept. Sweden's decision to block ChatGPT and Claude on government computers, while keeping Copilot, shows how quickly that trust question can turn into a practical restriction — read our report on Sweden blocking ChatGPT and Claude for the other side of the same coin.

What is the $895.4 billion commitment?

The single biggest number in the amendment is not revenue or loss. It is $895.4 billion: the total of Anthropic's purchase commitments and leases, substantially all of it for compute and infrastructure. Nearly half of that total, about $447.5 billion, is expected to be incurred before the end of 2030, and about $69.8 billion falls due in the 12 months after September 30, 2026. Reuters reported the filing on November 2, noting that the net loss had narrowed sharply from the second quarter even as the commitment total came into view.

Think of it as a very large, very long shopping list. Anthropic has promised to pay for data-centre capacity and computing power stretching years into the future, because training and running frontier models needs far more hardware than any one company owns outright. Against that, the filing reports $146.6 billion of liquidity — cash, short-term investments, and available credit — plus financing capacity from partners. That covers the next 12 months of commitments roughly twice over, before any IPO proceeds. The filing says Anthropic intends to fund the investments through the revenue it generates and the capital it raises. The honest translation: this plan only works if revenue keeps growing fast enough to meet bills that are already signed.

That is also why the filing landed in a nervous market. The last week of October brought a sharp sell-off in big technology stocks, as investors questioned whether enormous AI spending would ever pay for itself. Anthropic's filing is now the clearest test of that question: the fastest-growing revenue in the industry, attached to one of the largest spending commitments ever disclosed by a company about to list. If investors buy the story, it will be one of the biggest IPOs on record. If they do not, the valuation talk will move fast.

What valuation are investors talking about?

There is no valuation in the filing, because there is no price yet. Anyone quoting a number is quoting a report, a private-market trade, or an investor conversation — and those numbers have already moved. Anthropic was valued at $965 billion in a private funding round in May 2026. Reports since the public filing have put the possible IPO valuation between $1.5 trillion and $2 trillion, and Reuters reported on November 3 that some current and prospective investors are now weighing $1.5 trillion, down from roughly $2 trillion, after seeing the detailed financials at a recent investor event.

Reports have also put the possible amount raised at $100 billion or more. Again, that is reported, not filed, and it should be treated as a rumour with a good source rather than a fact. Until the amended filing carries a price range, there is no reliable way to calculate what Anthropic would be worth on day one. Our advice is blunt: ignore any social-media post that states the IPO price as fact this week. If it is not in an SEC filing, it is not settled.

One structural detail is already settled, and it matters for anyone thinking about influence rather than price. The filing sets up a share structure that keeps majority voting power with a company owned by the seven co-founders. Public shareholders would own the economics of the Class A shares trading as ANT, but the founders would keep control of the big decisions. That is increasingly common among founder-led technology listings, and it is exactly the kind of detail worth knowing before forming an opinion about the stock — this article is not investment advice, and no one should buy or avoid a share on the basis of a blog post alone.

How does Anthropic compare with OpenAI and Google?

Anthropic does not compete in a vacuum, and the last month has been a three-way sprint. Google launched Gemini 4 Carbon in late October, a cybersecurity-focused frontier model that tops the Artificial Analysis Intelligence Index — see our explainer on Gemini 4 Carbon — while OpenAI launched GPT-6.1 Terra, a fast computer-use model for ChatGPT Work and Codex, covered in our GPT-6.1 Terra explainer. OpenAI also confirmed that ChatGPT and Codex will share one usage allowance from December, a reminder that even the biggest labs are managing the same compute-cost pressure Anthropic discloses.

Cheaper rivals are part of the same pressure. DeepSeek's open-weight models and Kimi's K3.1 Pro keep showing that strong performance does not always need the most expensive infrastructure — read our coverage of DeepSeek V4.1 Flash and Kimi K3.1 Pro. Anthropic's answer, visible in this filing, is to spend its way to capacity and bet that Claude's quality and safety reputation hold premium demand. It is a credible bet, and an expensive one.

Lab	Latest flagship move (Oct–Nov 2026)	What we know for certain
Anthropic	IPO filing; Claude behind the filing	Q3 revenue $16.96B; ticker ANT planned
Google	Gemini 4 Carbon	Top Artificial Analysis score (59); limited defender-only access at launch
OpenAI	GPT-6.1 Terra; shared ChatGPT/Codex limit from December	Computer-use model in Work, Codex, and API
DeepSeek / Kimi	Open-weight and lower-cost challengers	Strong results at far lower reported cost

What is NOT confirmed: the leaked model names

You may have seen posts about "GPT-6.2 Astra" and "Claude 5.5" or "Fable 5.5" over the weekend. Neither OpenAI nor Anthropic has confirmed either name, and no independent benchmarks, prices, or release dates exist for them. The evidence so far amounts to a deleted catalogue entry and odd labels in usage logs. Treat those names as unconfirmed leaks, not launches, and do not make buying or building decisions around them until one of the companies says something on the record. Malay Mail's report on the filing captures the mood well: extraordinary growth, extraordinary spending, and a listing that will test how long investors' faith in the AI boom can last.

What does the IPO mean for Claude users?

Short answer: nothing announced, nothing immediate. The filing does not announce a Claude price rise, a feature change, or a new model. If you use Claude today, your plan, your limits, and your data arrangements do not change because a prospectus was filed. Anyone telling you otherwise is guessing.

Over a longer horizon, going public changes the incentives around a product. Public companies report every quarter, and every quarter invites the same question: is revenue growing, and are costs under control? That pressure can push a company toward higher-value enterprise deals, tighter free-tier limits, or more paid features — but it can equally push toward faster shipping and better tools to win the users who pay. Anthropic's own agent push points the second way: Claude Duet, Anthropic's AI agent, is exactly the kind of product that turns a chatbot into something businesses will pay for every month. Watch what happens to agent features, coding tools, and enterprise plans in 2027. That is where the filing's numbers will show up in your daily experience, if they show up at all.

The other thing to watch is trust. Anthropic's safety positioning is a selling point for schools, hospitals, law firms, and governments — the same customers who read a prospectus's risk section carefully. The filing reportedly devotes extensive space to risks, including AI systems behaving in unexpected ways. A public Anthropic will have to keep earning that trust in public, quarter after quarter, with regulators, customers, and now shareholders all watching.

How to read an IPO filing without getting lost

An S-1 is long, legalistic, and deliberately cautious. You do not need to read all of it to understand the story. Use these steps to get the useful 10% in about 20 minutes:

Start with the financial summary. Revenue, operating income or loss, and net loss, for the latest quarter and the same quarter last year. That is the growth story in four numbers.
Separate operating loss from net loss. The operating line shows the business. The net line often carries non-cash financing revaluations. If a headline loss looks absurd, check what sits between the two lines.
Find the commitments table. Purchase commitments and leases tell you what the company has already promised to spend, and when. Compare the next-12-months figure with cash on hand.
Read the risk factors as a to-do list for sceptics. Companies list what could go wrong, in their own lawyers' words. The most useful risks are the specific ones — customer concentration, compute supply, regulation — not the generic ones every filing carries.
Check what is blank. No price range, no share count, and no date means the deal is not priced yet. Any valuation you see before then comes from outside the filing.
Write your own one-paragraph summary before reading commentary. Drafting the summary yourself stops someone else's headline from doing your thinking for you. Our free AI Prompt Generator can turn "summarise this filing's revenue, losses, and commitments in plain English" into a sharp prompt for your AI assistant of choice — a practical way to interrogate a long document without skimming past the numbers that matter.

The honest risks

Pros — why investors are excited

Revenue growth few companies ever achieve: twelvefold year on year in Q3, with $33.24 billion in nine months.
Claude is a real, widely used product with enterprise traction, not a prototype looking for a market.
$146.6 billion of reported liquidity gives Anthropic room to meet near-term commitments before IPO proceeds arrive.
A safety-first reputation that appeals to the cautious, high-paying customers every lab wants.

Cons — why investors are nervous

Compute already eats 58% of revenue, and short-term capacity is squeezing margins further.
$895.4 billion in commitments must be paid whether or not revenue keeps growing at this pace.
The headline net loss, even if largely non-cash, shows how much financing complexity sits behind the growth.
Founder control, intense competition from OpenAI and Google, and regulatory scrutiny are all disclosed risks, not surprises — but they are real.

One detailed analysis of the amendment puts the tension neatly: the filing answers the two numbers investors most wanted — third-quarter revenue and total commitments — while still leaving price, share count, and deal size blank. Until those blanks fill in, this is a story about a business, not yet a story about a share price.

FAQs

When is the Anthropic IPO date?

No IPO date is set in the filing as of November 2, 2026. The roadshow can start no earlier than November 10, and reports point to a listing before Thanksgiving on November 26, but Anthropic has not confirmed a date.

What is Anthropic's stock ticker?

Anthropic plans to list its Class A shares on the Nasdaq Global Select Market under the ticker ANT. No shares can be bought until the registration statement becomes effective and trading begins.

How much revenue did Anthropic make?

Anthropic reported $16.96 billion in revenue for the three months ended September 30, 2026, up from $1.38 billion a year earlier. Revenue for the first nine months of 2026 was $33.24 billion.

Is Anthropic profitable?

No. Anthropic reported an operating loss of $567 million and a net loss of $23.29 billion in the third quarter of 2026. Much of the net loss reflects non-cash fair-value changes, so the operating loss is the cleaner measure of the business.

What is Anthropic's IPO valuation?

No valuation appears in the filing, because no price range has been set. Anthropic was privately valued at $965 billion in May 2026, and reports have discussed $1.5 trillion to $2 trillion, with some investors now weighing $1.5 trillion.

Will the IPO change Claude's price or features?

No change has been announced. The filing sets out no Claude price rise or feature change. Longer-term pressure to grow revenue and control compute costs could affect plans, but that is speculation, not news.

Our verdict

Anthropic's amended filing is the most revealing document the AI industry has produced this year, and it refuses to tell a simple story. The growth is real and extraordinary. The spending is real and extraordinary too. The $895.4 billion commitment is not a forecast — it is money already promised — and the entire investment case rests on revenue growing fast enough, for long enough, to meet it. That is a bet, not a certainty, and the filing is honest enough to show you both sides of it.

For readers, the useful takeaway is not a stock tip. It is a clearer picture of why AI feels both miraculous and expensive right now: the models are getting better because companies are buying computing power at a scale the industry has never attempted, and someone, eventually, has to pay for it. Whether Anthropic's answer — premium products, safety-led trust, and sheer scale — pays off is the question the roadshow will try to answer. We will update this article when the price range lands, because that is when the story stops being about a business and starts being about a share.

Last updated: November 3, 2026. Methodology: this article is based on Anthropic's amended S-1 filing (S-1/A, November 2, 2026) as reported by Reuters and Malay Mail, and on independent analyses of the filing linked above. Figures are as filed; 2026 interim figures are unaudited. Reported dates, valuations, and deal sizes are labelled as reported where the filing itself is silent. This article is not investment advice and is not sponsored by Anthropic or any other company mentioned. Questions or corrections? See our About page and Contact page.

Search This Blog

The AI Tools