Anthropic’s $2 trillion IPO question: How much growth can investors buy?

13:40, 23.10.26

More information has leaked from Anthropic's expected IPO prospectus, the document a company publishes before selling shares to the public. Investors are waiting for the official document because much of the AI company's financial information has been private. If Anthropic wants to go public in 2026, before its rival OpenAI, it is running out of time. Thanksgiving is coming, and Christmas will follow soon after. People connected to Anthropic have discussed starting a roadshow on November 9. A roadshow is a series of meetings where a company explains its business to possible investors. To meet that schedule, Anthropic would need to publish its prospectus very soon so investors have time to study the numbers before buying shares.

Some people think Anthropic is releasing its financial information little by little on purpose. This could help investors get used to the numbers before seeing the full report, which may be one of the year's most important financial documents. The idea is to prevent a big negative reaction when all the information becomes public. If that is the plan, it could be especially helpful for numbers that make the company look less successful.

A report from The Wall Street Journal this week gave the clearest picture yet of Anthropic's finances. Instead of showing just a few numbers, it explained more about the company's overall financial situation. Its revenue, meaning the money it brings in before expenses, was especially impressive. In the third quarter of 2026, Anthropic made $16.8 billion in revenue. That was 45% more than in the second quarter and even more than the $16.3 billion it made during the entire first half of 2026. By comparison, the company made just $1.1 billion in the first half of 2025.

These new figures help answer worries caused by an earlier leak that suggested Anthropic's growth was slowing down. Even if its percentage growth each month has become smaller, the company is still growing much faster than almost any ordinary business.

But the impressive revenue numbers also hide some risks. Anthropic depends heavily on a small number of customers. Just two customers provided 20% of its revenue in the first half of 2026. That could be a problem if one of those customers spends less money or switches to another AI provider. On the other hand, as more businesses start using AI in their daily work, Anthropic could gain many more customers and become less dependent on a few big ones.

The company's expenses are even more worrying. Investors already expect companies building the most advanced AI models to need enormous amounts of money. Even so, Anthropic's costs are surprisingly high. This is especially important because Anthropic representatives recently suggested that the company had made a profit in the previous quarter.

Anthropic reported a loss of $80.7 billion in the first half of 2026. About $72 billion of that was an accounting loss connected to debt that is expected to turn into company shares later. This means most of the reported loss was not money the company directly spent during that period. Reuters reported that Anthropic had a similar accounting loss of $42 billion the year before. The loss became larger because Anthropic's estimated value increased after another private fundraising round.

Even after removing that accounting loss, Anthropic still lost $8.7 billion. Most of its regular costs came from computing. In the first half of 2026, the company spent $9 billion on cloud computing services, computer chips and other computing needs. That was a huge increase from the $2.3 billion it spent on the same things during the first half of 2025.

And these costs may not show how much money Anthropic has promised to spend in the future. The company has signed contracts worth a total of $600 billion for AI computing infrastructure over the next ten years. It is also discussing more agreements worth another $150 billion. Google and Amazon are each reportedly connected to contracts worth $110 billion, while Microsoft has a smaller agreement worth $31 billion.

For Anthropic, these contracts mean it must keep spending enormous amounts of money, even if its growth slows or the business of AI becomes less profitable. But for the wider technology industry, these contracts are encouraging. Companies that make computer chips, provide cloud computing services or run data centers can point to these agreements as evidence that at least some of the huge investments in AI infrastructure are supported by real customer commitments.

This also shows how money can move in a circle within the AI industry. Investors who put money into companies that provide computing infrastructure have a reason to make sure big AI customers like Anthropic have enough money to pay their bills. Investors give money to AI companies, which then spend large amounts on computing services. That spending becomes revenue for other companies, some of which may be supported by the same investors. This system helps AI businesses grow quickly. But it also means the companies depend on one another, which could cause problems if investors start providing less money.

Anthropic's fast-growing revenue and increasing losses show something rarely seen on such a large scale. The company was founded only six years ago, yet it is on track to make about $60 billion in revenue this year. Large, well-known companies such as Unilever, drugmaker Eli Lilly, Cisco and Netflix needed years of building their businesses to reach that level of revenue.

But the same numbers that show how quickly Anthropic is growing also raise two big questions before a possible IPO. Can the company keep growing long enough to start making a profit? And can it convince investors that it is worth $2 trillion while it continues losing so much money?

The answers will partly depend on what happens in the wider economy and financial markets. Changes in bond yields, interest-rate decisions by the Federal Reserve and the possibility of more serious conflict in the Middle East could all make investors more or less willing to buy shares in an expensive technology company. These factors could matter even if Anthropic itself continues performing well.

If no major outside event disrupts the market, one useful comparison is SpaceX, the company led by Elon Musk. SpaceX sold shares to the public last summer in an offering that valued it at $1.77 trillion. Looking only at the financial numbers, Anthropic has a strong reason to compare itself with SpaceX. Even though Anthropic is losing money and is only six years old, it already makes more revenue than SpaceX and is growing faster.

SpaceX was also losing money in the quarter before its IPO. But according to financial results released yesterday, it returned to making a profit in its most recent quarter. This shows why investors sometimes accept large losses from a company when they believe it has a strong chance of growing much bigger in the future.

Still, there are reasons not to compare the two companies too closely. At first, investors struggled to accept SpaceX's high valuation. During its first few months on the stock market, its shares traded below the price investors had paid in the IPO. But after its latest financial results and discussions about a possible combination with Tesla, something people had expected for a long time, SpaceX's valuation rose above $2 trillion.

Investors had similar doubts about SpaceX's IPO and Google's IPO about 20 years ago. When Google went public, its valuation of more than $20 billion seemed extremely high. In both cases, investors had to decide whether the company's future possibilities were worth paying a price that looked too expensive by normal business standards.

Anthropic must also think about the value investors are placing on its rival OpenAI. OpenAI has delayed its own IPO for now. Instead, it is working on raising money from private investors at a reported valuation of $1.4 trillion. That number could influence how investors value Anthropic. But a price agreed on by private investors does not necessarily mean investors on the public stock market will be willing to pay the same amount.

In the end, Anthropic may face a common choice for companies preparing to sell shares to the public. Investment bankers often tell companies to "leave some money on the table." This means setting the IPO price low enough that new shareholders have a chance to make money if the share price rises after trading begins, instead of trying to get the highest possible price immediately.

If stock prices continue rising toward the end of the year, Anthropic might find it better to go public at a valuation of $1.5 trillion and give its shares room to rise. That could be more appealing than demanding a $2 trillion valuation and risking a disappointing start, with little or no increase in the share price. Anthropic has already grown faster than almost every normal measure would suggest possible. Its biggest challenge may not be showing investors that it is an important company. Instead, it may need to convince them that the profits it could make in the future will eventually be large enough to justify the enormous amount of money it is spending today.