What smart people are saying about Anthropic's leaked IPO prospectus numbers

• Anthropic's IPO prospectus is out, and the numbers are astonishing.
• For example, it spent $7.3 billion on computing and faces $518 billion in fresh commitments.
• Business Insider put together a list of what smart people are saying about those numbers.

Anthropic's IPO math is staggering.

The AI lab generated nearly $4.6 billion in revenue last year — about 12 times its 2024 total — while recording a $42 billion net loss, per an IPO prospectus seen by Reuters. The company spent $7.33 billion on compute and infrastructure in 2025 and has $518 billion in future cloud, computing, and infrastructure obligations.

The prospectus could value the company at $2 trillion, while the company is predicting an IPO after the November midterm elections, Reuters reported.

The report comes as Anthropic has been on a bit of a roller coaster ride. Employees, including CEO Dario Amodei, have warned that AI labs haven't secured a firm grasp on their autonomous AI systems amid reports of security breaches and hacking incidents. Anthropic dedicated 80 of the IPO's 261 pages to risk factors, per the Financial Times.

Meanwhile, last week, Anthropic released Claude Opus 5.5, its latest flagship model.

Here is what some smart techies and market watchers have said about the gobsmacking numbers. The company didn't immediately respond to a request for comment from Business Insider.

Lale Akoner, global market strategist at eToro

Akoner told Business Insider that the prospectus shows that Anthropic has a lot of work to do.

While the 2025 revenue growth shows how fast demand has accelerated, the "more meaningful figure is the $8.1 billion operating loss," she said, because the company is a "long way from proving it can turn rapid revenue growth into sustainable profits."

Still, if the IPO is successful, the effects could ripple through the market, Akoner said. She said it could set the stage for OpenAI's expected IPO and attract investment away from other tech companies.

"Ultimately, investors will need to decide whether Anthropic's growth can justify both its enormous valuation and its future spending commitments," she said.

Gary Marcus, professor emeritus of psychology at NYU

Marcus, a consistent AI skeptic, sees a perverse incentive.

"Anthropic's proposed valuation is easily calculated, as -50 times 2025 losses," he wrote on X. "The more they lose, the more they win!"

His post gets at the unease behind the huge numbers: Wall Street may reward companies for growth and infrastructure spending, well before they prove their businesses can generate consistent profits.

He also said that investors have to paper over concerns about rogue AI agents.

"You can only get there by largely ignoring all the ways things would go wrong," he wrote. "If I tell you that there is a 10% chance of annihilating humanity, the expected value is a loss of 800 million people, which would be by far the worst disaster in human history."

Ross Hendricks, an equity analyst at Porter & Company

Hendricks sees a bill hiding in Anthropic's fine print.

About $34 billion of the company's 2025 loss was a noncash accounting charge — meaning that amount did not leave the company's bank account last year. Instead, he said, the company could issue more shares, meaning early investors own a smaller slice of the company.

"The shareholder dilution on this will be something to behold," Hendricks wrote.

Jake Williams, cybersecurity risk expert

In a post on Bluesky, Williams zeroed in on another disclosure: nearly a quarter of Anthropic's 2025 revenue came from two customers.

"Siri, is it good when huge chunks of my revenue come from only two companies?" he wrote.