Anthropic Files IPO Registration Statement, a Full Comparison with OpenAI on Valuation, Revenue and Compute

Main text

Editor's note: Based on Anthropic's latest prospectus and recent media reports, this article provides a comprehensive comparison of the two companies' operating conditions and development strategies across four dimensions: valuation, revenue, profitability, and computing power investment. It aims to use intuitive data comparisons to provide investors with a reference for understanding the commercialization progress, competitive landscape, and future challenges of the two AI giants.

Cailian Press, October 30 — Artificial intelligence (AI) company Anthropic is accelerating its pace toward going public.

On October 26 local time, Anthropic publicly submitted a registration statement for an initial public offering (IPO) to the U.S. Securities and Exchange Commission (SEC), planning to list on Nasdaq under the ticker symbol ANT. According to media reports, the company could go public as early as mid-November, but the specific issue price and fundraising size have not yet been determined.

As one of OpenAI's main competitors, Anthropic has rapidly expanded its market share in recent years through the Claude series of models and the coding tool Claude Code, and has made significant progress in revenue scale and valuation.

Financial data disclosed in the latest prospectus show that Anthropic's revenue reached $11.555 billion in the second quarter of this year, surpassing OpenAI's approximately $6.7 billion in the same period, and it achieved quarterly operating profit for the first time.

As Anthropic is about to enter the public capital markets, the competition between the two AI giants over valuation, commercialization capability, profitability, and computing resources has also entered an important window for observation.

Since the beginning of this year, OpenAI and Anthropic have successively completed massive financing rounds, with valuations both approaching $1 trillion.

On March 31, OpenAI announced the completion of $122 billion in financing, with a post-money valuation of $852 billion. On May 28, Anthropic announced the completion of $65 billion in Series H financing, with a post-money valuation of $965 billion, more than double its $380 billion valuation in February this year and higher than OpenAI's most recent financing valuation.

However, OpenAI is seeking a higher valuation.

According to media reports on October 30 citing people familiar with the matter, OpenAI has lowered its new financing target from about $30 billion to about $20 billion, with a target valuation of about $1.25 trillion, lower than the $1.4 trillion previously proposed to investors.

The report noted that some investors have shown resistance to the higher valuation, and related financing demand is also less robust than before. As concerns about the sustainability of AI infrastructure investment heat up, some investors want to first observe Anthropic's IPO pricing before deciding whether to invest in OpenAI.

Anthropic's potential IPO valuation is also facing adjustment. According to media reports in October, some investors are considering giving the company a valuation of about $1.5 trillion, lower than the previous market expectation of about $2 trillion.

Aswath Damodaran, a professor at New York University's Stern School of Business and a valuation expert, estimated in an analysis published in October that Anthropic's equity value is about $1.475 trillion. He believes the company's rapid revenue growth supports the high valuation, but it will still need to continue expanding revenue in the future to absorb massive computing power investment.

Currently, Anthropic has not yet announced a formal issue price range.

As Anthropic takes the lead in advancing its listing, its issue pricing and post-listing market performance may become an important reference for investors assessing the valuations of OpenAI and other unlisted AI companies.

Revenue competition: Anthropic overtakes in the second quarter of this year

The revenue competition landscape between the two companies is also changing.

According to media reports in August citing data disclosed by OpenAI to investors, OpenAI's revenue was $5.7 billion in the first quarter of this year and rose to $6.7 billion in the second quarter, a quarter-on-quarter increase of about 18%.

In comparison, Anthropic's prospectus shows that the company's revenue was $4.731 billion in the first quarter and rapidly climbed to $11.555 billion in the second quarter, a quarter-on-quarter increase of about 144% and a year-on-year increase of 1,368%.

Based on the above data, Anthropic's second-quarter revenue was about 72% higher than OpenAI's.

Over a longer period, Anthropic's revenue growth has been particularly rapid. For the full year 2025, the company's revenue was $4.585 billion, while in the first half of 2026 alone its revenue reached $16.286 billion, equivalent to 3.55 times last year's full-year total.

According to media reports in October citing people familiar with the matter, Anthropic's third-quarter revenue further increased to $16.8 billion, up 45% quarter on quarter, slower than the approximately 144% growth in the second quarter.

Regarding the slowdown, Gil Luria, head of technology research at D.A. Davidson, believes that Anthropic's slower revenue growth is related to OpenAI strengthening competition in the enterprise market and enterprises adopting lower-priced open-weight AI models.

It should be noted that there are important differences in the revenue recognition standards of the two companies.

According to media reports on October 9, OpenAI's annualized revenue run rate as of the end of September was close to $50 billion, lower than the approximately $70 billion widely reported in the media.

This difference mainly stems from the way revenue from cloud channels is accounted for. When Anthropic sells Claude services through channels such as Amazon AWS and Google Cloud, it recognizes the related sales on a gross basis, whereas OpenAI uses a different net-basis accounting method for some channels.

Previously, investors tried to adjust OpenAI's annualized revenue to a gross basis more comparable to Anthropic's, which produced higher revenue estimates.

Separately, media reports around the same period said Anthropic's annualized revenue as of the end of July was about $65 billion. However, because the two companies' data correspond to different time periods and there are differences in accounting standards, this cannot be used directly to judge the latest revenue gap between the two sides.

Beyond revenue scale, the two companies' commercialization structures also differ.

Anthropic's prospectus shows that in the first half of this year, about 80% of its revenue came from usage-based services, with subscription revenue accounting for about 20%.

As of the end of September, Anthropic had more than 11,900 customers with revenue of more than $100,000 over the past 12 months, including more than 1,900 customers with revenue of more than $1 million, and about 250 customers contributed more than $10 million in revenue.

The spread of specialized tools such as Claude Code has driven Claude deeper into enterprises' software development and other workflows, becoming an important driver of Anthropic's revenue growth.

OpenAI, by contrast, has a broad product system covering both consumer and enterprise markets, including ChatGPT, Codex and API services. However, analysts point out that OpenAI must bear the cost of serving a large number of free ChatGPT users while facing increasingly intense price competition in the enterprise market.

Profitability divergence: Anthropic turns profitable for the quarter, while OpenAI remains deeply loss-making

As the revenue competition landscape changes, the two companies' operating results have also diverged markedly.

According to investor materials disclosed by the media, OpenAI posted an operating loss of $9.3 billion in the first quarter of this year, and the loss widened further to $12.3 billion in the second quarter, with an operating margin of about -184%. The report noted that the figure includes stock-based compensation expenses.

By comparison, Anthropic's prospectus shows that the company posted an operating loss of $1.667 billion in the first quarter, but achieved operating profit of $974 million in the second quarter, with an operating margin of about 8.4%, whereas it was still loss-making to the tune of $1.2 billion in the same period last year.

Anthropic's second-quarter operating profit is under US GAAP and likewise includes recognized stock-based compensation expenses. However, because OpenAI has not yet publicly disclosed complete financial statements, it remains difficult to confirm whether the two sides' other expense classifications and accounting treatments are fully consistent.

An important reason for the improvement in Anthropic's operating performance is that revenue grew significantly faster than computing costs.

The prospectus shows that the company's computing and infrastructure expenses in the second quarter were $5.558 billion, up about 300% year on year, significantly lower than the 1368% revenue growth in the same period.

Computing and infrastructure expenses as a share of revenue also fell to 48% from 177% in the same period last year, reflecting a significant decline in the relative burden of related computing expenses as revenue scale expanded.

However, Harrison Rolfes, a senior analyst at PitchBook, pointed out that some of Anthropic's contracted computing power has not yet been delivered for use, and the related expenses will be reflected gradually in the future. Therefore, the improvement in computing efficiency in the second quarter may not be sustainable, and subsequent cost increases could put pressure on margins.

On cash flow, Anthropic generated net cash inflow from operating activities of $3.929 billion in the first half of this year, compared with a net outflow of $1.513 billion in the same period last year. As of the end of June, its cash, cash equivalents and short-term investments totaled about $131.651 billion.

It should be noted that positive operating cash flow does not equal positive free cash flow; the company still has to bear substantial infrastructure investment and long-term contract payments.

OpenAI's future funding needs are likewise enormous.

According to media reports in September citing OpenAI investor presentation materials, the company expects cumulative revenue of $840 billion from 2026 to 2030, including full-year revenue of $350 billion in 2030.

But against the backdrop of large-scale investment in computing infrastructure, OpenAI expects a cumulative free cash flow shortfall of $278 billion over the same period. That forecast is narrower than the $305 billion in the May version, but still implies enormous external financing needs.

The report also noted that, based on cash burn forecasts at the time, the $122 billion OpenAI raised in March this year could be exhausted by 2028.

Overall, Anthropic has demonstrated quarterly operating profitability, while OpenAI still faces large operating losses and funding needs. However, as both sides continue to expand computing investment, long-term profitability remains under test.

Computing race: the long-term risks behind hundreds of billions of dollars in investment

Computing power is a key resource supporting both companies' model research and development and commercial expansion, and is also an important source of future funding needs.

According to media reports in September citing people familiar with the matter, as of the end of 2025, OpenAI had about 2GW of available computing capacity, while Anthropic had about 1.5GW.

However, both sides expect to reach about 5GW by the end of 2026 and further increase to about 10GW by the end of 2027. If the relevant plans are fulfilled, Anthropic will essentially catch up with OpenAI in available computing scale.

At the same time, the two companies are locking in long-term resources through different cooperation models.

OpenAI's most representative infrastructure project is the "Stargate" plan, and it continues to expand cooperation with Oracle, AWS, Microsoft and other companies.

According to media reports in July, OpenAI's previously signed data center capacity contract with Oracle reached about 6GW, most of which is still under development.

In addition, OpenAI's multi-year cloud computing contract with Amazon AWS has been expanded to $138 billion over eight years, including about 2GW of Trainium computing capacity.

OpenAI also plans to invest $20 billion to launch the Project Camellia data center in Georgia, with arrangements to gradually obtain 3.2GW of power supply from 2028 to 2032. This is also a representative project in which it has begun to participate more directly in the design and development of large data centers.

According to investor materials disclosed by the media in September, OpenAI expects computing and infrastructure spending of about $856 billion from 2026 to 2030, close to its projected cumulative revenue of $840 billion over the same period.

Anthropic, for its part, has adopted an infrastructure strategy involving multiple suppliers and multiple chip architectures, including Nvidia GPUs, Google TPUs, Amazon Trainium and AMD GPUs.

The prospectus discloses that Anthropic has committed to purchasing at least $114.5 billion of computing power from Google by 2033 and at least $110 billion of related services from Amazon by 2036. In addition, Anthropic has signed a procurement agreement with Microsoft worth at least $44.9 billion.

The company has also signed an equipment finance lease arrangement with Broadcom, with total payments over the relevant lease term of about $161.2 billion, excluding interest.

In addition, Anthropic has signed large-scale computing service agreements with companies such as SpaceXAI, Nscale and Lambda, further expanding its sources of computing resources.

For OpenAI and Anthropic, locking in computing resources in advance helps ensure model research and development and customer service capabilities, but if actual demand falls short of expectations in the future, massive long-term procurement and leasing arrangements could also turn into a financial burden.

At present, Anthropic shows strong revenue growth and improving operating performance, while OpenAI continues to expand its product lineup and long-term computing investment. The key to future competition between the two sides lies in whether they can convert massive computing resources and growing AI demand into sustainable profits and cash flow.

(Source: Cailian Press) Mid-article embedded advertisement