Anthropic Files S-1 to List on Nasdaq Under ANT…

[[**Anthropic plans to list on Nasdaq under the ticker ANT, with first-half revenue of $16.286 billion and a net loss of $80.748 billion; the loss stems mainly from accounting valuations, and the IPO will test whether AI commercialization can outpace compute spending.**

**Key points**

**1. The $80.7 billion loss is mainly due to accounting valuations**

About $72.388 billion of the net loss comes from fair value changes in convertible notes, derivative liabilities, SAFEs and other instruments, rather than cash burn, but infrastructure, personnel and marketing spending remain high.

**2. Compute contracts lock in both growth and cost risk**

Locking in AWS, Google, Nvidia and AMD compute in advance secures supply, but long-term procurement, leases and minimum payment obligations could squeeze margins if demand slows.

**3. Rapid revenue growth still needs to prove commercialization quality**

Revenue rose from $41 million in 2023 to $16.286 billion in the first half of 2026, but the key questions are embedding in enterprise workflows, customer stickiness, renewals and expansion, and gross margins.

**4. The IPO forces the market to reassess what kind of company an AI company is**

Anthropic combines software scalability, R&D risk and capital-intensive infrastructure characteristics, so pricing must assess growth, compute costs, contract flexibility and cash flow.

**5. Founder control protects the mission but needs governance oversight**

A public benefit corporation, a long-term benefit trust and special voting rights give the founding team significant control, which can protect long-term safety investment, but public shareholders have limited oversight influence. 2026-10-27 09:59 The largest IPO in history is also the AI company with the biggest loss: Is Anthropic dragging capital markets into a compute gamble? Dr. Xi Chunying's quick take

Introduction: Revenue grew 14-fold in half a year, yet the net loss exceeded $80 billion. What Anthropic's prospectus reveals is not just one company's books, but a turning point as the AI industry moves from a model race to a capital-efficiency race.

On October 26, 2026, Anthropic publicly filed its S-1 prospectus with the U.S. Securities and Exchange Commission, planning to list on Nasdaq under the ticker "ANT." This AI company, founded only five years ago, is challenging the record for the largest IPO in history. The prospectus discloses that Anthropic's revenue in the first half of this year reached $16.286 billion, but its net loss was even higher at $80.748 billion. The simultaneous appearance of rapid revenue growth and huge losses quickly made this listing the focus of global capital markets.

In my view, the most noteworthy thing about this prospectus is not the $80.7 billion loss itself, but what it reveals about the changes taking place in the AI industry: frontier AI is no longer a pure software business, but a heavy-asset race driven jointly by models, compute, capital and governance structures. The market is willing to pay high valuations for technological growth, but whether a company can turn growth into cash flow ultimately depends on whether it can effectively control compute costs and improve returns on capital.

The real test for Anthropic's IPO is not "does AI have a market," but "can AI commercialization outpace compute spending."

1. The $80.7 billion net loss cannot simply be understood as cash burned

The prospectus shows that Anthropic's revenue in the first half of 2026 was $16.286 billion, compared with about $1.163 billion in the same period last year; its operating loss in the same period was $693 million, while its net loss reached $80.748 billion. The huge gap between the two loss figures is mainly because about $72.388 billion came from fair value changes in convertible notes, derivative liabilities, SAFEs and other financing instruments.

The related accounting valuation changes were recorded in current profit and loss, significantly amplifying the book net loss. Therefore, the $80.748 billion net loss cannot be directly equated with the company burning $80.748 billion in cash in half a year. But "mainly accounting factors" does not mean operating costs can be ignored. In the same period, compute and infrastructure spending reached $8.945 billion, personnel costs were $3.152 billion, and sales, marketing and partner costs were $3.168 billion, indicating that the company's main business is still in a high-investment phase.

When looking at the financial performance of AI companies, one should not focus only on net profit, nor pick only the better-looking operating loss figure. Net profit is affected by changes in the valuation of financing instruments, operating loss reflects the accounting result of the main business, and cash flow relates to whether the company can actually support future investment; these three sets of accounts must be viewed together.

The real question to ask is: When can Anthropic stably generate operating cash flow? How much additional compute and infrastructure investment is needed for each additional dollar of revenue? Can the commercial value brought by improved model capabilities continue to exceed R&D, inference and capital expenditure?

2. Compute contracts are a ticket to growth, and may also become a cost shackle in the future

Anthropic's revenue growth follows a clear chain: improved model capabilities, increased enterprise applications, expanded customer usage, and revenue growth. But AI services differ fundamentally from traditional software: every model response requires computing resources, and the more users there are and the more complex the tasks, the greater the compute consumption.

The prospectus discloses that Anthropic uses different hardware including AWS Trainium, Google TPU, Nvidia GPUs and AMD GPUs, and schedules compute from multiple suppliers in a unified way to improve price and performance. The company has signed large-scale compute procurement, leasing and infrastructure arrangements to reserve resources for model R&D and customer service.

From a growth perspective, locking in compute capacity in advance can prevent supply shortages from constraining business expansion; from a financial perspective, long-term contracts can also become rigid costs. If compute is insufficient, the company cannot keep up with customer demand; if too much compute is locked in, once demand growth slows, idle resources and minimum payment obligations may compress profit margins.

Market reports add up Anthropic's various arrangements to hundreds of billions of dollars and summarize a considerable portion of them as "pay whether you use it or not." But different contracts contain procurement commitments, leases, minimum payment obligations, contract caps and different terms, and cannot simply be treated as the same kind of debt. What should really be examined is payment timing, minimum usage volumes, exit clauses, price adjustment mechanisms, and whether the company's actual utilization rate matches revenue growth.

My judgment is that the future winners among AI companies will depend not only on who has the strongest models, but more on who can turn compute into a sustainable unit economics model. If revenue growth must be supported by equally or even faster-growing compute investment, greater scale does not necessarily mean higher capital efficiency. Only when model efficiency improves, hardware costs fall and customer value grows at the same time can high growth possibly translate into genuine operating leverage.

III. Moving from a model company to enterprise workflows is Anthropic's real commercial test

Revenue grew from $41 million in 2023 to $386 million in 2024, $4.585 billion in 2025, and then $16.286 billion in the first half of 2026. Anthropic's commercialization speed is very prominent.

But rapid revenue growth is only the first step. Whether AI products can enter enterprises' core workflows determines whether customers are trying them briefly or purchasing continuously over the long term. If a model only answers questions, the switching cost may not be high; if it can be embedded in code development, legal review, customer service operations, research analysis and internal knowledge management, customers are more likely to keep using it and expand deployment.

Anthropic's opportunity is to develop from a model supplier into an enterprise work platform; its challenge is that model capabilities do not automatically translate into pricing power. Cloud platforms, enterprise software vendors, developer tools and customers' internal technology teams may all participate in the deployment of AI products and the distribution of value. Models iterate quickly, and today's technological lead does not guarantee the same commercial advantage in the future.

Therefore, investors need to watch not only revenue growth, but also large-customer concentration, channel costs, customer renewals and expansion, changes in gross margin, and how much revenue and cash contribution each unit of compute can create. Revenue growth can prove that market demand is forming, while the quality of growth depends on customer stickiness, cost control and profit conversion.

IV. Founder control can protect the long-term mission, but it also needs to withstand governance scrutiny

Anthropic operates as a public benefit corporation and has set up a long-term benefit trust and special voting rights arrangements. Relevant disclosures show that the founding team has significant voting control, and the long-term benefit trust participates in oversight of the company's long-term mission. Media summarize this arrangement as the founding team controlling about 50.1% of voting rights.

This design has a real logic. Frontier AI R&D cycles are long, and safety investment may conflict with short-term profit goals. If management is entirely driven by quarterly performance pressure, the company may be more willing to pursue rapid releases and commercial expansion while compressing safety research and risk management.

But after going public, public shareholders also have to face the other side of highly concentrated control: they invest capital but may not have commensurate governance influence. Founder control can protect long-term strategy, but it may also weaken external shareholders' constraints on capital allocation and major decisions. Whether the public benefit mission can ultimately be implemented depends not only on declarations in the corporate charter, but also on how the board oversees, how the trust exercises its power, and who makes decisions when the mission conflicts with commercial interests.

Long-termism is not immunity from accountability. A truly mature governance structure must both protect R&D and safety investment from short-term pressure and ensure that those who hold control accept transparent and effective oversight.

V. This IPO is in fact repricing the AI industry chain

The significance of Anthropic's listing application is not only that it may set a new IPO fundraising record, but also that it forces investors to reassess the commercial attributes of AI companies. Anthropic simultaneously has the expansion capacity of a software platform, the R&D risks of a technology company, and the capital-intensive characteristics of an infrastructure company. Valuing it only as an asset-light software company may underestimate the long-term obligations brought by compute contracts; valuing it only as an asset-heavy company may ignore the expansion potential of models and platforms.

The capital market ultimately has to answer several questions: How long can high growth last? Can compute costs fall? Will long-term contracts limit operational flexibility? Can corporate governance protect public shareholders? And can revenue scale translate into stable cash flow?

For small and medium-sized listed companies in Hong Kong, Anthropic's story also offers direct lessons. AI cannot be just a hot label in financing roadshows; companies must clearly explain which business link AI has changed, how much cost reduction, efficiency improvement or new revenue it has brought, and at the same time explain how the investment will be recovered, how compute and data costs will be controlled, and how operating results will be verified.

The largest IPO in history is also the most loss-making AI company in history. These two labels reveal not a simple contradiction, but that the AI industry is entering a new stage: technological imagination is already enough to attract huge capital, and the next competition will revolve around capital efficiency—whoever can use less compute to create more sustainable revenue may be able to move from technological leadership to commercial leadership. This is true for Anthropic, true for the entire AI industry, and true for every company that hopes to use AI to reshape its business.

Has published 62 articles on Huxiu. This content is authorized by the author for publication, and the views represent only the author and not Huxiu's position. 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 in the Huxiu APP[ ]