With enthusiasm for the potential of artificial intelligence (AI) still strong, a few heavyweights in the space are preparing to go public, including Anthropic, the creator of the popular Claude large language model. There's a chance that the market will put a $2 trillion market cap on Anthropic when it IPOs, in which case it will edge out Space Exploration Technologies (SpaceX) as the largest IPO in history.
Whether Anthropic will actually reach that $2 trillion valuation remains to be seen, but its recently released IPO prospectus offers interested investors a more detailed look at the state of its business.
Image source: Getty Images.
Anthropic is spending a lot on compute
Anthropic's revenue increased by 1,050% to $4.6 billion in 2025. The more eye-popping number, though, was its $42 billion in net losses. The good(ish) news is that around $34 billion of that loss came from an accounting charge. Anthropic's operating loss (a figure that only reflects the profits and losses that relate to running the business) was around $8 billion.
For tech companies in the high-growth stage that Anthropic is in, the combination of fast-growing revenues and mounting bottom-line losses isn't unusual. The company is still early enough in its lifecycle that investors will stomach it, but "all burn, no profit" only gets so much grace for a company asking for a $2 trillion market cap.
Much of Anthropic's spending right now is going toward buying the compute power needed to train and scale Claude and other infrastructure. In 2025, it spent $7.3 billion on this, up from $400 million in 2023 and $2.5 billion in 2024. That accounted for 58% of its $12.65 billion in total operating expenses in 2025.
One of Anthropic's biggest challenges right now
Every company's IPO prospectus includes a risk section. Anthropic's mentions that one of its challenges right now is ensuring that autonomous AI models remain contained. However, it highlighted a much larger issue: the possibility that advanced AI could pose "catastrophic or existential risks to humanity."
Anthropic acknowledges that there's a real chance that AI systems could begin putting their interests above human interests. Whether that means manipulating information, resisting human actions to shut down poorly performing or dangerous AIs, or overall deception, that's not quite an encouraging note from one of the most influential developers of AI tools.
The difficulties that Anthropic is having with balancing the long-standing tech sector approach of "move fast and figure it out as you go" with operational methods suitable to preserving the safety of humanity aren't unique to it. Sam Altman, CEO of rival AI giant OpenAI, has recently begun encouraging the companies that are developing AI to slow down.
Either way, Anthropic's IPO will draw a lot of investor attention. It looks like it will be the first frontier AI lab to test the public markets, absorbing all the scrutiny that comes with that role.
