Anthropic's IPO: What Its S-1 Reveals About The AI Boom

One of the most eagerly awaited events in the generative AI market is finally approaching. Anthropic has filed to become a publicly traded company. For the first time, investors can get a detailed look at the finances behind one of the world's leading artificial intelligence companies.

On October 26, 2026, the company behind Claude filed its S-1 registration statement with the Securities and Exchange Commission. The filing shows extremely fast revenue growth, huge commitments to infrastructure spending, and financial risks that were mostly hidden before. Anthropic could be valued at nearly $2 trillion, making its public offering one of the largest in stock market history.

Investors can now start answering important questions about AI's rapid growth and whether its business model can last. The numbers are striking. Anthropic earned $16.3 billion in revenue during the first half of 2026, reported a net loss of $80.7 billion, and revealed huge financial commitments to the computing infrastructure needed to build and run its AI models.

For years, investors have put enormous amounts of money into leading AI companies, betting that these businesses will grow dramatically in the future. Now, Wall Street can examine how much money they actually earn, how much it costs to provide their services, and whether their financial results support valuations approaching $2 trillion.

OpenAI is also preparing for a possible public offering. Anthropic's financial disclosures could therefore set a standard for an entire generation of AI companies. Can businesses built around rapidly improving AI models really be worth trillions of dollars? Or are investors expecting a future that may be much more expensive, competitive, and difficult to turn into profits than they realize? Could the AI market be a bubble that will eventually burst, or can these businesses make money in a lasting way?

Anthropic's Financial Results Tell Two Very Different Stories

Anthropic's revenue growth is remarkable. The company earned just $386 million in 2024, followed by about $4.6 billion in 2025. Revenue reached $16.3 billion in the first six months of 2026, almost 14 times the amount earned during the same period a year earlier, according to the company's filing.

However, its reported net loss of $80.7 billion needs a closer look. About $72.4 billion of that loss came from accounting adjustments that did not involve actual cash spending. These adjustments were largely linked to financial agreements that can turn into company shares and became more valuable as Anthropic's valuation rose.

In other words, Anthropic previously raised money through financial agreements that can eventually be exchanged for company shares. As the company's value increased, the potential value of these agreements to their holders increased too. Accounting rules required Anthropic to record that increase as an expense, even though it did not actually spend that money to run its business.

The increase in value is not entirely theoretical because these agreements can eventually turn into shares, reducing the percentage of the company owned by existing shareholders. Still, the huge reported loss makes Anthropic's operating performance look much worse than its underlying financial results suggest.

This does not mean Anthropic actually spent $80.7 billion running its business. However, these accounting losses reflect the rising cost of financial obligations connected to its increasing valuation. The situation shows that raising money at ever-higher valuations can have significant financial consequences, including reducing existing shareholders' ownership or creating obligations that must be settled in the future.

Anthropic's actual operating loss for the first half of 2026 was $693 million, down from $2.1 billion a year earlier. More importantly, the company earned $974 million in operating profit during the second quarter alone, after reporting an operating loss of $1.7 billion in the previous quarter.

This improvement suggests that Anthropic is starting to benefit from economies of scale. In other words, it is earning more money without its costs rising at the same rate. Revenue in the second quarter reached $11.6 billion, up from $4.7 billion in the first quarter, while computing expenses did not increase by the same proportion.

Put another way, Anthropic's revenue rose about 144% from the first quarter to the second quarter of 2026, while computing and infrastructure costs rose by only about 65%. This is a meaningful improvement in how efficiently the company operates and could be a good sign for investors looking for a business model that can last.

Of course, one profitable quarter does not prove that the company can remain profitable. Anthropic's costs could jump when new computing capacity becomes available. It may have to pay these costs well before the related customer revenue arrives.

Even so, better AI economics could allow revenue to grow faster than the cost of developing and delivering more advanced models. The key question is whether Anthropic can keep doing this consistently.

Can Anthropic Justify a $2 Trillion Valuation?

People have discussed a possible public valuation for Anthropic of between $1.5 trillion and $2 trillion, although the final offering price has not been set. At a valuation of $1.5 trillion, the company would be worth about 33 times its second-quarter annualized revenue of $46.2 billion. At $2 trillion, that figure would rise to nearly 43 times annualized revenue.

The Information reported that Anthropic earned about $16.8 billion in revenue during the third quarter. That equals an annualized revenue rate of $67.2 billion if the same pace continued for a full year. Using this newer figure, the valuation would be about 22 times annualized revenue at $1.5 trillion, or 30 times at $2 trillion. These are still high valuations, but they may be easier to defend. Even so, they assume a great deal from a business with a limited history of operating profits.

For comparison, an October 26 market analysis estimated that Nvidia, Microsoft, and Alphabet were valued at about 17, 12, and 9 times their past 12 months of revenue, respectively. These companies also have longer records of making profits and generating cash.

Investors are effectively betting that Anthropic can keep growing revenue at an exceptional rate, build lasting relationships with customers, and earn a meaningful share of the economic value created by its AI technology.

The company's customer numbers offer some support for this view. Anthropic reported more than 11,900 customers each generating at least $100,000 in annual revenue. About 250 customers were each spending $10 million or more.

However, the pace of growth is already slowing. Third-quarter revenue rose about 45% from the previous quarter, compared with growth of more than 140% in the second quarter. That is still very fast growth, but the slowdown raises questions about how long Anthropic can keep growing quickly enough to justify its high valuation.

Several factors could make future revenue growth harder to maintain. Businesses are paying closer attention to the financial returns they get from AI and the cost of using AI tokens, which are small pieces of text or other data that AI models process. They are increasingly asking whether they need expensive, advanced AI models for every task.

At the same time, cheaper open-weight models, especially those from Chinese companies, and a growing number of smaller models designed for specific tasks, including new decision-making models, are giving customers more choices beyond general-purpose advanced AI systems.

AI companies may lose some of their ability to charge high prices if competing models become very similar and customers can easily switch providers based on cost, performance, or availability. This would make it much harder to maintain high profit margins. Competition could put pressure on prices and revenue growth even if overall AI use continues to rise.

More AI use does not automatically mean more revenue, either. Reuters Breakingviews reported that Anthropic's token use rose about 90% between early July and late September, while the related spending increased by only 19%, according to data from Ramp. This suggests that greater AI use does not always lead to a similar increase in revenue.

Anthropic also acknowledged in its S-1 filing that about 5% to 7% of its revenue during the first nine months of 2026 came from customers later found to be breaking its usage policies or business terms. Some of this activity involves what Anthropic says is unauthorized model distillation. This happens when AI developers use Claude's outputs to train their own competing models. Anthropic is working harder to stop this activity, even if doing so means giving up some revenue. The company expects revenue from customers breaking its rules to fall, but says the effect on growth is large enough to matter and remains uncertain.

Details About Data Center and Computing Commitments

According to Reuters, Anthropic has agreed to spending arrangements worth more than $500 billion with major technology partners over roughly the next decade. This does not mean it must pay the entire amount immediately or that all of it is an unconditional debt. The agreements have different terms, including minimum purchase requirements, leases, and payments that depend on how much computing capacity it uses. Even so, they create substantial financial risks.

Anthropic must arrange access to enormous amounts of computing power before it knows how much demand it will receive. Many business customers, meanwhile, pay for AI capacity based on how much they actually use instead of promising to buy a fixed amount over many years. This creates a potentially costly and risky mismatch. Anthropic's infrastructure commitments may last for years even if customer demand changes, competitors lower their prices, or newer technologies make computing more efficient.

The company has considerable financial resources. At the end of June 2026, its balance sheet showed about $131.7 billion in cash, cash equivalents, and short-term investments. Even with this much money available, investors need to distinguish between the cash Anthropic currently has and the much larger financial commitments tied to its expansion plans.

Another concern is Anthropic's relationships with Amazon, Google, and Microsoft. These companies provide its infrastructure, help distribute its products, invest in it, and sometimes compete with it. This creates a complicated relationship in which companies can be both partners and rivals.

Anthropic disclosed that about 48% of its revenue in the first half of 2026 came through cloud marketplaces. Its five largest customers accounted for about one-quarter of its revenue during the first nine months of the year.

The situation is complicated. Some companies helping finance Anthropic's growth are also selling it the infrastructure it needs to grow. At the same time, they are building their own competing products. Investors will need to work out how much of Anthropic's business comes from steady, growing demand from end customers and how much depends on a so-called circular investment cycle, in which connected technology companies invest in, buy from, and sell to one another.

What Anthropic's IPO Means for OpenAI

Companies in the generative AI market often move in step. When one company releases an updated model, its competitors soon follow. That makes it important to consider what could happen when OpenAI eventually goes public.

OpenAI confidentially filed for a U.S. IPO in June 2026, but it has not publicly released an S-1 filing with the same level of detail. Reuters reported in October that OpenAI was seeking about $30 billion in additional private funding at a possible valuation of more than $1.4 trillion.

There is less public information about OpenAI's finances than there is about Anthropic's. However, Reuters reported that OpenAI told investors in October that its annualized revenue rate was about $50 billion. Anthropic's reported third-quarter annualized rate was about $67 billion. These figures may not be directly comparable, however.

As Axios explained, Anthropic and OpenAI record some sales made through cloud partners differently. This makes their reported revenue figures difficult to compare directly. For example, suppose a customer spends $100 using an AI model through a cloud provider that keeps $20. Anthropic would report the full $100 as revenue and record the $20 as an expense. For a similar transaction, OpenAI would report only the $80 it receives.

Both companies would keep the same $80 before paying other expenses. But Anthropic would report 25% more revenue from the transaction. Neither accounting method is necessarily wrong. However, this difference can make Anthropic's reported revenue look larger than OpenAI's compared with the underlying business activity.

The companies also have different business strategies. OpenAI benefits from ChatGPT's strong brand, large consumer audience, developer network, and growing business customer base. Anthropic has built a strong position in business AI and software development, including through its Claude Code product.

OpenAI's public financial disclosures will need to show whether its larger consumer business can make attractive profits after paying for computing, product development, distribution, and infrastructure. For investors, a well-known brand and a large number of users do not replace the need for profitability.

Anthropic's IPO could force investors to compare the two companies more directly, especially if public markets give Claude a very different valuation from the one private investors currently give ChatGPT.

AI's Next Test Is Profitability and Sustainable Finances

There is strong evidence that demand for AI is real. Anthropic's fast revenue growth and growing number of business customers make it difficult to argue that advanced AI is only speculative technology with no real commercial success. Even the early signs that Anthropic may be able to make a profit are encouraging for people wondering whether the AI boom will collapse.

But real demand does not automatically justify the prices investors are willing to pay. The risk of a bubble comes from a possible gap between the revenue AI companies can earn in the future and the enormous amount of money they need to spend to generate it.

The industry is building huge amounts of computing capacity because it expects demand to keep growing. AI companies expect better models to create new uses for the technology and encourage customers to spend more. Infrastructure providers expect AI companies to keep buying chips, networking equipment, and computing services. Each part of this investment cycle depends heavily on what the other parts are expected to do.

If businesses adopt AI more slowly than expected, prices for AI models fall faster than costs, or computing infrastructure goes unused, these expectations could prove wrong.

Anthropic's disclosures show that advanced AI can generate enormous revenue and that its operating performance can improve steadily. However, they do not yet prove that these businesses can consistently earn returns that justify trillion-dollar valuations and extraordinary long-term spending commitments. The basic risks facing the market remain.

This article was originally published on Forbes.com