The Zhitong Finance App learned that artificial intelligence startup Anthropic officially disclosed the US initial public offering (IPO) documents. While showing rapid revenue growth, it also unraveled the huge costs, long-term computing power expenditure commitments, and corporate governance arrangements behind the development of cutting-edge AI models. This high-profile listing will not only test investors' valuations of Anthropic, but may also become an important yardstick for Wall Street to measure cutting-edge AI business models.
Anthropic secretly submitted an IPO application in June 2026 and disclosed the relevant documents on October 26. The company plans to list on the NASDAQ under the ticker symbol “ANT,” with Goldman Sachs and Morgan Stanley leading the underwriting team. The document did not disclose the number of shares to be issued and the proposed issue price range.
As the developer of the Claude series of AI models, Anthropic has rapidly expanded its business scale in recent years due to growing demand from corporate customers, but its development has also been accompanied by high computing and infrastructure expenses. As market concerns about AI data center investment returns, financing costs, and industry valuations heat up, Anthropic's current IPO will be an important opportunity for investors to test the future of AI commercialization.
Revenue grew rapidly, and there was a significant differentiation between operating loss and net loss
According to the prospectus, Anthropic's revenue increased from 41 million US dollars in 2023 to 386 million US dollars in 2024, and reached 4.59 billion US dollars in 2025. In the first half of 2026, the company's revenue further increased to approximately US$16.29 billion.
In the three months ending June 30, 2026, Anthropic's revenue increased 1368% year over year, reflecting the continued growth in demand from enterprise customers for its AI models and related services.
However, revenue expansion does not mean that the company is already profitable. According to documents, Anthropic's operating loss in 2024 was US$2.98 billion, expanding to US$8.06 billion in 2025; operating loss for the first half of 2026 was US$693 million.
Meanwhile, the company's net loss for the first half of 2026 reached US$80.7 billion. Earlier reports indicated that the sharp increase in net losses was mainly related to changes in the fair value of convertible bonds. Therefore, the huge gap between net loss and operating loss cannot simply be understood as the company's operating activities directly consumed US$80.7 billion in cash in the first half of the year.
Anthropic's revenue mainly comes from pay-as-you-go services for enterprise customers. Customers pay based on the amount of information processed, that is, the number of so-called “tokens” and the type of model used. In addition, the company also earns revenue through paid subscription products such as Pro and Max.
This pay-as-you-go model enables Anthropic to increase revenue as the frequency and scale of the use of AI models by enterprise customers increases. The company disclosed that its net revenue retention rate reached 900%, indicating a significant increase in the expenses of existing customers compared to the previous period. In the third quarter, 79% of the company's revenue came from customers other than the top five, indicating that its revenue sources were not entirely dependent on the few largest customers.
However, Anthropic also indicated in the prospectus that revenue and accounts receivable are still concentrated on a limited number of customers and distribution channels. As the scale of the business expands, customer concentration, revenue continuity, and whether enterprise customers can continue to increase AI spending are still issues investors need to pay attention to.
Over $500 billion in spending promises, computing power costs have become a key test
For Anthropic, computational power is the foundation that drives model development and commercialization, and is one of the company's biggest sources of cost.
Its computing and infrastructure expenses include computing fees paid to vendors such as Amazon Cloud Services (AWS) and Google Cloud, as well as expenses related to data center facilities, rental computing capacity, equipment rental, and energy. For the past three years, Anthropic's computing and infrastructure expenses have been far higher than revenue.
The prospectus revealed that Anthropic has promised to spend more than $500 billion on six partners over the next ten years. These promises are an important part of its large-scale AI infrastructure construction plans, but they also mean that the company will face huge capital requirements in the future.
Specifically, Anthropic promised to pay at least 110 billion US dollars to Amazon for cloud computing in the next seven to 10 years, about 114.5 billion US dollars to Google, and 44.9 billion US dollars to Microsoft for cloud computing services.
Some of the agreements are highly binding: Anthropic said many of the terms in the agreement are irrevocable or require the company to pay the relevant payments regardless of actual usage.
This kind of long-term commitment helps Anthropic lock in computing resources ahead of time to guarantee the training and deployment of more advanced models, but may also limit the company's cost adjustment space. If demand for AI falls short of expectations, increases in model efficiency change computing power requirements, or commercialization returns of related services fall short of expectations, companies may still need to bear a considerable amount of fixed expenses.
It should be noted that the above figures are expenditure commitments of a different nature that span many years. They are not equivalent to Anthropic's current debt, nor do they mean that all of the relevant payments have been made. What really affects the company's financial performance will be how these promises are converted into actual expenses in the future, and whether related computing power can generate sufficient revenue and profit.
Amazon and Google are both investors and cloud service providers
Anthropic's capital and computing power system also shows a characteristic of the current AI industry that has received widespread attention: some major investors are also key suppliers to the company.
Amazon has invested at least $13 billion in Anthropic and agreed to invest up to $20 billion more. Google, on the other hand, promised to invest up to $40 billion in Anthropic. At the same time, Anthropic has promised to pay huge cloud computing fees to Amazon and Google for years to come.
This relationship helps Anthropic obtain capital support and infrastructure resources, and allows cloud service providers to benefit from growing demand for AI models. However, at the same time, capital investment, cloud service procurement, and computing power construction are intertwined, drawing market attention to the AI industry's “circular investment” structure.
Under this structure, tech giants not only support the development of AI companies through investment, but also receive revenue through cloud service contracts. For investors, the key question is whether these funds and transactions can ultimately bring sufficient external customer demand, continuous cash flow, and reasonable return on capital, rather than simply driving the expansion of capital flows and expenditure within the industry chain.
Currently, the market is increasingly concerned about the speed of return on AI infrastructure investments. As investment in the construction of data centers, chips and computing resources continues to increase, AI industry chain enterprises may face higher financing needs and valuation pressure if related revenue is not realized in a timely manner.
Valuation expectations cool down, and IPOs will test AI companies' pricing logic
Anthropic's IPO comes at a time when the market is re-examining the AI investment boom. Recently, global stock markets have fluctuated markedly, and investors are beginning to pay more attention to whether AI infrastructure spending driven by debt financing can generate sufficient returns within a reasonable period of time. Higher borrowing costs and higher bond yields have also increased market caution about overvalued technology stocks.
Earlier reports said that after Anthropic disclosed some financial information, the company's valuation expectations of some current and potential investors dropped from about $2 trillion to about $1.5 trillion.
This does not mean that Anthropic's business growth has been reversed, but rather reflects that investors, in addition to focusing on revenue growth and market share, are also paying more and more attention to capital expenses, long-term contract constraints, profit paths, and actual returns on capital investment when evaluating AI companies.
Anthropic's valuation reached $965 billion in the May 2026 round of financing. As the company prepares to enter the open market, its final issuance valuation will provide investors with a more direct reference to measure the growth potential and financial risk of cutting-edge AI model companies.
Competitor OpenAI also secretly submitted an IPO application in June, and the market anticipates that it may go public in early 2027. Meanwhile, companies such as Google and Elon Musk's SpaceX are also investing huge sums of money to develop models, build data centers, and promote the layout of dedicated AI chips.
As a result, Anthropic's listing is not only related to the financing of one company, but may also affect the market's valuation judgment of other large AI companies. Its open market performance will help investors assess how high valuations can be supported by the rapid spread of AI models, and how long it will take for huge computing power investments to be transformed into sustainable profitability.
The structure of public service companies, the founder's control, and governance arrangements are also worth paying attention to
In addition to financial performance and cost commitments, Anthropic's corporate governance structure is also an important part of the disclosure of the prospectus.
Anthropic was founded in 2021 by a group of former OpenAI researchers including brother and sister Dario Amodei (Dario Amodei) and Daniela Amodei (Daniela Amodei). Dario is the CEO, and Daniela is the president and chairman of the board of directors.
Anthropic uses a Public Benefit Corporation (PBC) structure. This means that while pursuing returns to investors, the company also has a legal obligation to balance the public interest. The company said it will maintain this structure even after listing, and an independent body will elect a majority of board members.
However, the company plans to continue to let the founding team and an independent trust play a key role in governance. According to the disclosure of documents, most of the seven co-founders will hold a total of 50.1% voting rights on key corporate matters through a Class F share. Related matters include some director elections and other issues submitted for shareholders' vote.
This arrangement means that after Anthropic's listing, public shareholders' influence on the company's decisions may be limited. The company believes that centralized control will help it reduce related risks while advancing the development of AI technology; however, for open market investors, how to balance founder control, board independence, and shareholders' rights is still an important factor in evaluating corporate governance.
AI security risks have become another long-term test
Anthropic's current IPO is also facing additional scrutiny due to AI security issues. The company's own research found that in controlled tests, advanced AI systems may show acts such as code destruction, assisted fraud, and information manipulation.
The prospectus also warns that the model may present so-called “self-protective behavior,” including trying to resist closure, concealing, or manipulating information, and even showing acts similar to extortion.
Anthropic CEO Dario Amodei previously called on the global AI industry to slow down the release of some new capabilities to address related risks. However, at the same time, the company still needs to continue to advance model research and development in order to stay ahead in the competitive AI market.
Anthropic points out in the document that the company's success depends largely on continuing to develop the most advanced and capable AI models; failure to do so may adversely affect the business.
This also highlights the dilemma Anthropic faces: on the one hand, companies need to continuously improve their model capabilities to maintain enterprise customer needs and market competitiveness; on the other hand, the stronger the model capabilities, the more complex the security, governance, and regulatory challenges.
As Anthropic enters the open market, investors need to evaluate not only whether its revenue growth and computing power expenses can generate reasonable commercial returns, but also whether the company can continue to advance technology research and development in the midst of fierce competition while managing AI security risks.
Anthropic's IPO will bring these issues to the open market. Its final valuation and post-listing transaction performance may become an important reference for the capital market to determine the commercial value, capital requirements, and long-term risk of cutting-edge AI companies.