Anthropic seeks IPO at a $2 trillion valuation, while investor expectations have been lowered to $1.5 trillion
Anthropic is seeking to go public at a $2 trillion valuation and plans to raise about $100 billion, but at investor briefing days some investors are already measuring it at a $1.5 trillion valuation. The company's third-quarter revenue growth slowed to 45% quarter-over-quarter from 147% in the second quarter, with quarterly net new revenue falling from about $6.9 billion to about $5.2 billion; SemiAnalysis estimates its annualized revenue at about $74 billion. Anthropic disclosed at least $518 billion in cloud service, computing power, and infrastructure spending commitments over the next decade with six partners, about 80% of which is non-cancelable.
Why read this
AI score 64
The original article presents the disagreement over Anthropic's IPO pricing along three lines: revenue growth, computing spending commitments, and investor valuation expectations.
Topics
Body · AI translation
[[**Anthropic is seeking an IPO at a $2 trillion valuation, but the market has already pushed the price down to $1.5 trillion, mainly because revenue growth plunged from 147% in Q2 to 45% in Q3, and the market's enthusiasm for high premiums on AI star companies is fading.**
**Key Points**
**1. Valuation cut by the market**
Anthropic is seeking to list at a $2 trillion valuation and plans to raise $100 billion, but at an investor communication day, some investors had already lowered their valuation expectations to $1.5 trillion.
**2. Growth slows sharply**
The company's Q3 revenue growth slowed from 147% quarter-over-quarter in Q2 to 45%, with quarterly net new revenue falling from about $6.9 billion to $5.2 billion; annualized revenue is about $74 billion, below some investors' earlier optimistic expectations of more than $100 billion.
**3. Intensifying competition and distillation controls**
Besides being attributed to tighter distillation controls (related revenue accounts for about 5%-7% of revenue), the more direct blow to the slowdown comes from OpenAI's Codex coding tool and cheaper Chinese open-source models, forcing Anthropic to cut prices in response.
**4. Massive computing spending under pressure**
Anthropic's spending commitments for computing power and other items over the next decade total at least $518 billion, of which Google and Amazon each account for more than $110 billion; as of the end of last year it had only $20.28 billion in cash, and even the $125 billion raised cumulatively is still hard to cover costs, making an IPO almost the only option. 2026-10-16 19:00 Is Anthropic worth $2 trillion? Shanshang Quick Look
This article is from the WeChat public account: Shanshang, author: Shanshang team, original title: "Is Anthropic worth $2 trillion?"
Barring surprises, we may see the largest IPO in history by the end of the year.
Multiple media outlets report that Anthropic is seeking to list at a $2 trillion valuation and plans to raise about $100 billion. By comparison, the IPO valuation Musk sought in June this year after merging SpaceX and xAI was only $1.77 trillion.
But as the listing date gets closer, the market is beginning to question Anthropic's ultra-high valuation. At a recent investor communication day held by Anthropic, some investors instead measured Anthropic at a valuation level of $1.5 trillion.
The direct trigger is that Anthropic's growth is no longer there. Even though Anthropic disclosed at the investor communication day that the company's revenue in the third quarter of this year still grew as high as 45% quarter-over-quarter, with quarterly revenue more than three times that of the whole of last year, it has fallen sharply from the 147% growth in the second quarter, disappointing some investors.
Wall Street has high hopes for Anthropic's listing path. To some extent, Anthropic's IPO directly reflects whether the entire AI industry is prospering. If the listing goes poorly, it will also affect other companies' listing prospects. Previously, OpenAI had postponed its IPO to next year.
Anthropic is still continuously releasing good news to build confidence for the upcoming IPO. They disclosed at the investor communication day that the company achieved adjusted profitability in the third quarter. According to previous media reports, Anthropic's full-year operating loss last year exceeded $8 billion.
As of now, Anthropic has completed nearly 20 rounds of financing, with cumulative financing amounting to as much as $125 billion. In May this year, Anthropic announced the completion of a $65 billion Series H round, with a post-investment valuation of $965 billion.
Growth has already slowed
For a startup founded only 5 years ago, Anthropic's growth can be called a miracle.
Last year, Anthropic's full-year revenue was only $4.6 billion. But in the first quarter of this year alone, quarterly revenue reached $4.7 billion, exceeding the whole of last year. In the second quarter, Anthropic's revenue further increased to $11.6 billion, with quarter-over-quarter growth as high as 147%. In half a year, the company's revenue has reached more than three times that of the whole of last year.
The explosive popularity of Claude Code played an indispensable role. Anthropic officially opened Claude Code to the public in May last year, focusing on coding scenarios, and it quickly became popular from the end of last year.
In February this year, Anthropic disclosed that Claude Code's annualized revenue had exceeded $2.5 billion, more than double that at the start of the year. Among this, more growth came from B-end customers, with the number of enterprise subscriptions increasing to four times that at the start of the year, and enterprise usage contributing more than half of revenue.
Claude Code
The number of large customers is also increasing rapidly. In February this year, there were more than 500 enterprise customers with annualized spending of more than $1 million at Anthropic, and by early April there were more than 1,000. Anthropic's annualized revenue also jumped from about $9 billion at the end of last year to more than $30 billion in early April, and further broke through $47 billion in early May.
At that time, Silicon Valley tech companies were in the midst of the Token maxxing trend, with many tech companies including Microsoft, Meta, and Amazon running internal Token rankings to encourage employees to use AI more to complete work. But at the same time, such measures also bred a large amount of unnecessary computing consumption, with some employees' personal monthly Token consumption reaching as high as more than 280 billion.
Anthropic benefited a great deal from this, and at one point even faced a computing power shortage. To meet demand, Anthropic successively signed large computing power agreements with Amazon, Google, Broadcom, Microsoft and Nvidia, and partnered with Fluidstack to build data centers. In May of this year, Anthropic also booked all the computing power of SpaceX's Colossus 1 data center.
But entering the third quarter, Anthropic's growth slowed markedly, with quarter-on-quarter revenue growth falling to about 45%, and single-quarter new revenue dropping from about $6.9 billion in the second quarter to about $5.2 billion.
Anthropic's annualized revenue growth also appears to be falling short of market expectations. Some investors had previously optimistically estimated that Anthropic's annualized revenue by year-end could far exceed $100 billion. But according to estimates by research firm SemiAnalysis, Anthropic's current annualized revenue is about $74 billion.
Anthropic said part of the reason for the revenue decline is that they tightened controls on distillation. Previously, Anthropic had been accusing other AI companies of distilling its models, but these distillations also brought considerable revenue. According to the Financial Times, Anthropic disclosed that distillation-related revenue accounts for about 5%–7% of revenue.
Blaming distillation alone still seems insufficient to explain Anthropic's sharply falling quarterly growth. Some market analysts believe that competitive pressure from OpenAI and a large number of Chinese open-source models may be the direct cause of Anthropic's slowdown.
Since the beginning of the year, OpenAI has continued to update its models and Codex programming tool, diverting Anthropic's fastest-growing programming market. At the same time, cheaper Chinese open-source models have further increased price pressure, forcing Anthropic and OpenAI to cut prices in response.
SemiAnalysis analyst Joey Brookhart told the Financial Times that over the past three months, investors have been downgrading expectations for Anthropic's annualized revenue. "Now Anthropic has fallen into a two-horse race, rather than standing alone as it did in the first half of the year."
Even larger computing power spending
Unlike in the past, when Anthropic's management always emphasized the dangers and hidden risks of AI on various occasions, at this investor meeting, Anthropic CEO Dario Amodei and CFO Krishna Rao focused on the company's operations and profitability.
In addition to disclosing that the company achieved adjusted profitability in the third quarter, Anthropic also achieved positive cash flow. For every $1 they invest in computing power, they can generate $1.7 in revenue. Excluding stock-based compensation expenses, the company's adjusted operating margin for the quarter was 11%.
A year ago, Anthropic's revenue was still far from enough to cover expenses. The prospectus obtained by Reuters shows that Anthropic's revenue last year was about $4.6 billion, operating expenses were about $12.65 billion, and operating losses exceeded $8 billion. Among them, computing power and infrastructure spending was about $7.3 billion, higher than all revenue that year.
Computing power is a cost black hole that no AI company can avoid. To maintain technological leadership, large-model companies have to invest huge computing power training costs, and as user scale and invocation volume grow, inference expenses also continue to increase. Among the domestically listed Zhipu and Minimax, more than 70% of R&D investment is spent on computing power.
Anthropic's computing power spending is even greater. According to Reuters, Anthropic disclosed in its prospectus that over the next decade, its cloud service, computing power and infrastructure spending commitments with six partners will reach at least $518 billion. About 80% of this is non-cancelable, or must be paid whether or not it is used.
Google and Amazon account for the bulk of it. Anthropic has committed to paying Google at least $111.1 billion between April 2026 and July 2033, and Amazon $110 billion between May 2026 and April 2036. Under the contracts, even if actual usage spending is insufficient, the company still needs to make up the difference.
The agreement with Microsoft involves $31.4 billion, with the payment period running from November this year to May 2033. In addition, Anthropic also bears about $161.2 billion in equipment leasing obligations related to Broadcom, most of which are likewise non-cancelable.
Anthropic tried to explain to investors the necessity of huge computing power spending. Anthropic said access to computing power is becoming a key constraint on AI development, and future demand for advanced AI systems is likely to exceed available supply, and will be "mainly limited by the availability of computing power."
The cumulative approximately $125 billion in funding Anthropic has obtained through financing is still insufficient to cover its almost astronomical computing power bills. Anthropic's prospectus shows that as of December 31 last year, Anthropic's total cash, cash equivalents and short-term investments were only $20.28 billion.
After a record $65 billion Series H financing, if Anthropic wants more funding support, an IPO is almost the only option. According to previous media reports, Anthropic plans to raise up to $100 billion, which would make it the largest IPO in global history.
But for Anthropic, whose growth is currently slowing, the market no longer seems willing to give a higher premium. Channel: Finance and Economics. This content is published with the author's authorization; the views represent only the author and do not represent the position of Huxiu. If you have objections or complaints about this manuscript, please contact tougao@huxiu.com. Everyone who is changing and wants to change the world is on the Huxiu APP[ ]