Thanks to Anthropic’s soaring valuation, its prospectus shows blow-out losses, due to how the tech giants’ investments in the AI firm were structured.
A leak of Anthropic’s prospectus reveals that the AI giant made $4.6 billion, and lost $8 billion on an operating basis last year. But the total of its losses last year was a whopping $42 billion.
That would have been one of the worst annual losses on record for an American company —almost matching the $29.1 billion loss reported by Citigroup in the 2008 financial crisis. But Anthropic didn’t blow this money on compute, talent or takeovers. Instead, it’s a bookkeeping charge measured by Generally Accepted Accounting Principles (GAAP).
The charge is largely due to two major Anthropic investors’ stakes being marked up. Google first invested in the company in mid-2023, but later that year it would invest $2 billion in the form of debt known as convertible notes. These can later be swapped for shares in the AI startup. Amazon also invested $8 billion in late 2024 at a $46 billion valuation, similarly via convertible debt.
“The loss is a consequence of good news.”
Joshua Ronen, an accounting professor at the NYU Stern School of Business
Now, with the company targeting a reported $2 trillion valuation for its IPO, the value of the shares Google and Amazon were promised have ballooned, too. Anthropic needs to mark that down as an accounting loss.
It’s not clear if the $34 billion in losses similarly accounts for other investors.
“The loss is a consequence of good news,” Joshua Ronen, an accounting professor at the NYU Stern School of Business, wrote in an email to Forbes. “The charge is enormous precisely because the valuation soared.”
Google noted in its last earning statement that it held $124.3 billion in private startups. The search giant didn’t disclose the name of these investments, but noted that this stemmed from largely one company. The size of its startup investments is so vast that it created an accounting blip for even a company of Google’s scale. In July, it recorded $99 billion in “other income” that stemmed from its startup investments in SpaceX and “another private startup.” The New York Times reported in March 2025 on court filings showing that Google owned 14% of Anthropic.
Amazon also noted in its July earnings report that its net income had leapt to $62.6 billion up from $18.2 billion last year, largely thanks to its investment in Anthropic. That surge in income was largely attributable to a $53.4 billion mark-up on its stake in the startup. Amazon has said it has invested $18 billion into Anthropic, and has the right to invest $20 billion more.
Both Amazon and Google have continued to invest in subsequent rounds for Anthropic which has raised over $126 billion, according to Pitchbook. Some of Amazon and Google’s note have already been converted to shares.
Anthropic still burned a hefty amount of cash operating its business: developing its Claude models, paying salaries and buying compute power. But $8 billion is nothing compared to the $518 billion it plans to spend on AI infrastructure over the next decade. Google’s cloud division and Amazon’s AWS arm are likely to be some of the biggest beneficiaries of this spending bonanza.
Anthropic, Google and Amazon didn’t immediately respond to requests for comment.
Anthropic’s market debut is highly anticipated, after the San Francisco-based startup confidentially filed its prospectus in June — a milestone event that would also shape the IPO environment for its archrival OpenAI. It also comes as the wider public has been debating the existential safety of AI, and what regulations could serve as guardrails, as AI models from the world’s biggest AI labs have hacked third parties.
Anthropic CEO Dario Amodei was one of a handful of top tech executives that yesterday signed an agreement with President Donald Trump at the White House to self-police the development of what the president has branded “Super Intelligence”. Earlier this month, CEO Dario Amodei called for a slowdown of frontier AI development, calling for “prudence” as the labs advance their models and amid a spat of hacking incidents involving AI agents.
It’s not the only drama that has consumed the company. Earlier this year, Amodei engaged in a high-profile standoff with the Department of War over the Pentagon’s ability to use Anthropic’s AI for domestic mass surveillance and autonomous weapons. After the tense back-and-forth, the government labeled Anthropic a “supply chain risk” — a devastating blow to any company that does business with the military. Anthropic subsequently sued the DoW, and a judge put an injunction on the designation.
Want to see more Forbes articles on your feed? Tap here to make Forbes Australia a preferred source on Google.
Look back on the week that was with hand-picked articles from Australia and around the world. Sign up to the Forbes Australia newsletter here or become a member here.
