Anthropic IPO: Is the $1.4 Trillion Claude Maker Overvalued or Ahead of Its Time?

The Anthropic IPO is the biggest stock market debut ever attempted. The company behind the Claude AI models plans to list on Nasdaq under the ticker ANT, with pricing expected on 18 November and trading from 19 November 2026, at a valuation of about $1.4 trillion. It arrives in the most nervous week the AI trade has seen all year, just days after Jensen Huang resigned as Nvidia’s CEO. So is this the top of an AI bubble, or the first chance for ordinary investors to own a piece of the next Microsoft? This guide walks through the numbers in plain language, with charts, so a first-time investor can make up their own mind.

Key takeaways

• Anthropic is selling 825 million shares at $70 to $78 each, raising up to $64.4 billion, plus a $26 billion private placement from its chip and cloud suppliers. Orders are reported at about six times the shares on offer.
• Revenue went from $41 million in 2023 to $4.6 billion in 2025 and $33.2 billion in the first nine months of 2026. The October 2026 annualised run-rate was about $81 billion.
• The scary “$104 billion loss” is mostly a non-cash accounting charge. The real operating loss for nine months was about $1.26 billion, and the business was operating-profitable in Q2 2026.
• The real risk is the bill for computing power: about $895 billion of commitments over the coming decade, with roughly $70 billion due in the next 12 months.
• At $1.4 trillion the stock trades at about 17 times its run-rate revenue. That is not crazy for a company growing more than tenfold a year, but it leaves little room for disappointment if growth slows.
• Jensen Huang left Nvidia over personal and security questions, not because AI demand collapsed. Still, the AI trade is shaky, and this IPO will be its biggest test.

Anthropic IPO at a glance

All company figures in this article come from Anthropic’s Form S-1 filed with the US SEC and its amendments, unless stated otherwise. Reports from the Financial Times say strong demand has led bankers to consider pricing above the range, which could lift the valuation past $1.5 trillion.

First, what does Anthropic actually sell?

Think of Anthropic as an AI brain for rent. Anyone can chat with Claude for free or for a monthly subscription, but most of the money comes from businesses. Companies plug Claude into their own software through an API (a pay-per-use connection) to write code, read contracts, answer customer queries and run multi-step tasks on their own. Software developers use a product called Claude Code to write and fix code. Every time a business sends work to Claude, Anthropic earns a small fee per unit of text, called a token.

FREE AND SELF-PACED

Open Varsity: 13 free courses for traders and investors

From stock market basics to options strategies, quant trading and taxation, built around Indian market examples.

• 407 chapters, beginner to expert
• Options, Python, AmiBroker AFL and quant tracks
• Risk management and trading psychology

Start learning free →by Marketcalls

That pay-per-use model is why revenue can grow so fast: once a large company wires Claude into its daily workflow, its bill rises with usage, without a new sales pitch. If you want to understand how these models are compared, our guide to LLM metrics and benchmarks explains the scorecards, and our explainer on graph engineering and AI agents shows how companies chain many AI agents together, which is what drives the heavy usage.

Investor mood right now: excited and nervous at the same time

The mood around the Anthropic IPO is split. On one side, demand for the shares is enormous: the offer is reported to be about six times oversubscribed, and some of the world’s biggest chip and cloud companies are buying $26 billion of stock alongside the public. On the other side, the wider AI trade has been bruised:

• The valuation came down. In September an earlier draft filing pointed to about $2 trillion. Talks then slipped to $1.5 trillion and then to $1.1 to $1.3 trillion before settling at the current range.
• The US midterm results hit AI stocks. Democrats won both houses of Congress, raising fears of curbs on new data centres. AI hardware stocks sold off hard in early November, with Korea and Taiwan hit the most.
• Nvidia lost its founder. Nvidia shares fell about 12.75% in the week of 9 to 13 November after Jensen Huang’s resignation, while rivals AMD and Broadcom each gained about 18%.
• Growth slowed a little. Anthropic’s Q3 2026 revenue grew 47% over Q2, compared with 144% in Q2, and the company slipped back to a small operating loss.

Interestingly, money that left AI hardware markets has been flowing into non-AI markets such as India, a story we covered in our Nifty momentum analysis. Investors are not abandoning AI. They are becoming picky about which part of AI they pay for.

Is the AI bubble collapsing after Jensen Huang left Nvidia?

Not because of Huang, at least not yet. It is important to be precise about what happened. Huang resigned as CEO and board member on 12 November 2026 after reports about his personal relationships and an independent board investigation into whether he had been targeted by Chinese intelligence. Nvidia has not given a reason, and its board says it has no reason to believe national security was compromised. CFO Colette Kress is interim CEO (Al Jazeera). His exit was about governance and personal conduct, not about a fall in demand for AI chips.

That distinction matters for investors. A bubble bursts when buyers stop paying. So far, the people who pay for AI are still spending heavily:

So the honest answer is that the AI market looks less like a bubble that is bursting and more like a market that is repricing. The warning sign to watch is circular money: chip makers and cloud companies investing in AI labs that then spend that money on the same chips and cloud. We wrote about the risks in AI bubble and cyber security risk. Nvidia’s earnings on 17 November, two days before Anthropic starts trading, will be the next big read on demand.

Inside the Anthropic S-1: the numbers that matter

1. Revenue is growing faster than almost any company in history

Anthropic made $41 million in 2023, $386 million in 2024 and $4.6 billion in 2025. In just the first nine months of 2026 it made $33.2 billion, and the company says its annualised revenue run-rate reached about $81 billion in October, up from $71 billion in September. A run-rate is simply the latest month’s revenue multiplied by 12. It is a useful speed gauge, but it is not the same as a full year of actual sales.

2. Revenue finally overtook the computing bill

For a company like Anthropic, the biggest cost is renting computing power: the chips, data centres and electricity that run Claude. In early 2025 Anthropic spent $2.51 on compute for every $1 it earned. By Q2 2026 it was spending just $0.48. That is the single most important trend in the filing, because it shows the business gets more efficient as it grows.

Calculations from the filing put gross margin at roughly 40% in 2025 and about 85% in the first half of 2026. For comparison, good software companies run gross margins of 70% to 80%. If that holds, Anthropic would look much more like a software company than a capital-hungry hardware business.

3. Businesses keep spending more every year

The number of customers paying Anthropic more than $100,000 a year rose from about 20 at the end of 2023 to more than 11,900 by September 2026. Over 250 customers spend more than $10 million a year. Analysts estimate net revenue retention at around 900%, which means a typical customer that started a year ago now pays about nine times as much. No single customer is more than 10% of revenue, though the top five together make up about a quarter.

4. The “loss” headline is mostly an accounting charge

Headlines screamed that Anthropic lost $104 billion in nine months. That number is real under accounting rules, but most of it is not cash leaving the company. When Anthropic raised money in the past through convertible notes and similar instruments, those were promised future shares. As the company’s value shot up, the accounting value of those promises rose too, and that rise is recorded as a loss. In the first half of 2026, $72.4 billion of the $80.7 billion net loss came from this one non-cash item.

The better number to watch is operating income, which measures whether the core business makes money. On an adjusted basis Anthropic lost $4.9 billion in 2025 but made $1.6 billion in the first half of 2026. Reported operating income was a $974 million profit in Q2 2026, then a $567 million loss in Q3. Nine-month operating loss was about $1.26 billion, small compared with $33 billion of revenue.

The $895 billion compute bill: the biggest risk in the Anthropic IPO

To keep up with demand, Anthropic has signed commitments for computing power and infrastructure worth about $895 billion over the coming years, with roughly $70 billion due within the next 12 months. The seven biggest named deals add up to about $637 billion, led by Broadcom equipment leases, SpaceXAI, Google and Amazon.

Two details stand out. First, the Nscale and Lambda deals, together about $86 billion, are for renting Nvidia chips, so Nvidia’s fortunes and Anthropic’s are closely linked. Second, Anthropic told investors that its cloud partners increasingly see it as a competitor and have not provided all the computing capacity it asked for. Google, Amazon and Microsoft are at the same time Anthropic’s landlords, investors and rivals. That is why Anthropic has spread its bets across seven suppliers.

Here is how to think about it as a beginner. Imagine a restaurant that has signed a 10-year lease for ten new kitchens because the queue outside is long. If customers keep coming, it is a brilliant move. If the queue shrinks, the rent is still due. About $430 billion of Anthropic’s named commitments are minimum or fixed payments that must be paid whether or not the demand shows up. That is the bet investors are really buying: that AI usage keeps growing fast enough to fill those kitchens.

Is the Anthropic IPO overvalued or futuristic?

The simplest way to judge a fast-growing company is the price-to-sales ratio: how many dollars investors pay for each dollar of annual revenue. At $1.4 trillion and an $81 billion run-rate, Anthropic would trade at about 17 times sales. That is similar to fast-growing software companies such as Snowflake and Datadog, even though Anthropic is growing many times faster.

The chart above flips the question. Growth always slows eventually, and mature tech companies usually trade at 5 to 8 times sales. For a $1.4 trillion valuation to hold up once Anthropic matures, it would need roughly $175 billion to $280 billion of yearly revenue. That is two to three and a half times today’s run-rate. Given that revenue grew more than tenfold in the last year, reaching that within a few years is possible. It is not guaranteed.

Our view: the Anthropic IPO is priced for a very good future, not for a perfect one. It is not cheap, but it is no longer the $2 trillion dream it was in September. Whether it proves futuristic or overvalued depends almost entirely on one thing: whether revenue keeps growing faster than the compute bill.

What you are not buying: control

Beginners often miss this. Buying ANT shares gives you a share of the profits, but almost no say in how the company is run:

• Class A shares (what the public buys) carry votes, but a single Class F share held by an entity owned by the seven co-founders controls 50.1% of the vote on most matters.
• Class N shares, bought by the strategic investors, have no votes at all.
• Anthropic is a public benefit corporation, which means it is legally allowed to weigh AI safety and the public good alongside profit. A Long-Term Benefit Trust of independent trustees elects a majority of the board.
• Insiders face lock-up rules: directors, officers and employees can sell up to 10% of their holdings at listing, others must wait 180 days or longer. Expect selling pressure when those windows open.

AI in the next 5 to 10 years: what could happen

Nobody can predict a decade of technology, but the big trends are visible today. Here is what most industry forecasts point to, and what each would mean for Anthropic:

• AI moves from chatting to doing. The next wave is AI agents that complete whole tasks: closing the books, writing and testing software, handling insurance claims. Agents use many times more computing than a chat, so revenue per customer could keep rising.
• The cost of intelligence keeps falling. Each new generation of chips and models makes the same answer cheaper. That helps margins, but it also lets competitors offer similar quality for less, so price wars are likely.
• Power becomes the bottleneck. The International Energy Agency expects global data centre electricity demand to more than double to about 945 TWh by 2030, with AI data centres alone more than quadrupling. The winners will be companies that secured power and chips early, which is exactly what Anthropic’s huge commitments are trying to do.
• Big Tech keeps spending. AI capex is running above $800 billion in 2026 and companies such as Amazon and Alphabet already guide higher for 2027.
• Regulation tightens. Data centre limits, AI safety rules and chip export controls will shape who can grow how fast. Anthropic’s safety-first image could be a strength here.
• Consolidation. Many industries end up with two or three dominant players. Today’s frontier AI leaders are Anthropic, OpenAI and Google. Not all of them will keep their position for ten years.

For a long-term investor the key question is not “Will AI matter in ten years?” It almost certainly will. The question is “Will Anthropic still be one of the two or three leaders, and will it make enough profit to justify today’s price?” Those are very different bets.

How a new investor should approach the Anthropic IPO

• Do not chase day one. Hot IPOs often jump on listing and then drift. SpaceX soared after its June 2026 debut and then gave back those gains. Waiting for the first quarterly results as a listed company costs little.
• Size it small. Treat ANT as a high-risk growth stock. Keep it a small slice of a diversified portfolio so a 40% fall would hurt, not ruin you.
• Watch three numbers every quarter: revenue growth, compute expense as a share of revenue, and operating income. If revenue growth keeps beating compute growth, the story is on track.
• Mark the lock-up dates. Around 180 days after listing, insiders can sell more shares. Prices often weaken ahead of that date.
• For Indian investors: ANT will trade in US dollars on Nasdaq. You can buy it through international investing platforms under the RBI’s Liberalised Remittance Scheme, but remember currency risk, foreign tax rules and higher brokerage on US trades.

Frequently Asked Questions

When is the Anthropic IPO?

Pricing is expected on 18 November 2026 and the shares are expected to start trading on Nasdaq under the ticker ANT on 19 November 2026.

What is the Anthropic IPO price and valuation?

The price range is $70 to $78 per share, valuing Anthropic at about $0.92 to $1.02 trillion on basic shares and up to about $1.4 trillion fully diluted. Reports say strong demand could push the final price above the range.

Is Anthropic profitable?

Partly. It made an adjusted operating profit of $1.6 billion in the first half of 2026 and a reported operating profit in Q2, but slipped to a $567 million operating loss in Q3. Its large reported net loss is mostly a non-cash accounting charge.

Did Jensen Huang leave Nvidia because the AI bubble burst?

No. Huang resigned over personal conduct and a board investigation into possible security concerns. AI demand indicators such as Big Tech capex and Anthropic’s own revenue are still rising, although AI stock prices have become more volatile.

Is the Anthropic IPO overvalued?

At about 17 times run-rate sales it is expensive but not extreme for a company growing more than tenfold a year. To justify the price as growth slows, revenue would likely need to reach $175 billion to $280 billion a year. The biggest risk is its $895 billion compute commitment if demand falls short.

MEMBERSHIP

Marketcalls Elite Traders Club

Every Marketcalls course and live mentorship in one membership: 127 courses and 2,500+ hours of learning.

• Market Profile, Orderflow, AFL, Pine Script and Python
• Live mentorship and TradEdge trading room
• Premium tools and the Elite Slack community

View membership plans →by Marketcalls