Details of Anthropic's much-anticipated IPO prospectus showed its revenue rose roughly twelvefold last year to nearly $4.6 billion, though the maker of the Claude AI models lost more than $8 billion from operations over the same period, and it has committed to spending at least $518 billion on computing infrastructure over the next decade.
The revelations from its confidential IPO prospectus, reviewed and reported initially by Reuters, offer investors their first peek at the finances of a company whose listing could value it at more than $2 trillion.
It also shows the degree to which Anthropic is going all in on future spending, as about 80% of the $518 billion it's committed must be paid regardless of how much computing power the company actually uses.
As per Reuters and other outlets, Anthropic posted a net loss of about $42 billion in 2025. Roughly $34 billion of that was an accounting charge reflecting a higher estimated value for financing arrangements that could eventually convert into Anthropic stock.
Excluding writedowns tied mostly to earlier fundraising, the San Francisco company's operating loss topped $8 billion. Total operating expenses reached $12.65 billion, with compute and infrastructure accounting for $7.33 billion of that. That's roughly triple the 2024 level, and probably a telling sign of its desire to remain competitive amid a broader stepped-up arms race in AI.
All told, Anthropic ended 2025 with $20.28 billion in cash, cash equivalents and short-term investments.
The prospectus also flags a point of concentration risk, as two customers generated nearly a quarter of the company's 2025 revenue. Beyond that, the company disclosed that many of its largest clients have no long-term contracts and could reduce or stop spending.
The buildout spending plan laid out in the prospectus spans six partners, and most of it cannot be undone.
"If our actual spend falls short, we must pay Google the difference," the company said in the filing. It added that similar terms apply to the Amazon agreement. The Microsoft commitment can be canceled only if Microsoft commits an uncured material breach.
Anthropic also carries about $161.2 billion in Broadcom-related equipment lease obligations. Neither side can cancel them except in the event of a default.
Among the more modest and flexible commitments, agreements with Elon Musk's xAI could lead to as much as $84.5 billion in spending on Nvidia-based computing capacity through 2029, but most of that can be canceled with 90 days' notice. Advanced Micro Devices has agreed to buy up to $5 billion of Anthropic stock and to supply computing capacity expected to exceed $20 billion.
In making the case for its aggressive buildout bet, Anthropic argued that computing power has become the bottleneck for the industry. The company said future demand for advanced AI will be "limited principally by the availability of compute."
The filing is direct about the conflicts in Anthropic's key relationships: Amazon, Google and Microsoft each act as investor, customer, cloud provider, distributor and competitor at once. Anthropic said those overlapping roles create incentives that "may not be fully aligned" with its own.
"If the compute we have access to from third parties is curtailed, repriced, or terminated ... our business, financial condition, and results of operations could be adversely affected," the company said.
In response, Anthropic is moving away from a cloud-only model toward dedicated data centers and directly leased chips.
The company also faces questions about its own technology. In controlled tests, its researchers have found that increasingly autonomous AI models can behave in unexpected and potentially harmful ways, including sabotaging code and assisting fraud. Earlier this month, Chief Executive Dario Amodei penned a lengthy think piece-slash-manifesto urging the industry to slow the release of new capabilities. Even so, Anthropic launched its Opus 5.5 model last week to compete with OpenAI's GPT-6 Astra.
Despite being one of the frontrunners in the AI race, Anthropic has also established itself as a conscientious objector, at one point clashing with the White House over how its tools are used. That dispute led the Pentagon to temporarily blacklist the company, a move a federal judge blocked last month.
Early in the summer, Anthropic confidentially submitted a draft registration statement to the SEC, and the document has not been made public.
Reuters, citing unnamed sources, has reported that the offering is likely to come after the November midterm elections. Anthropic declined to comment.
A valuation above $2 trillion would be more than double the roughly $965 billion Anthropic was estimated to be worth in May. It would also top SpaceX's record-setting June market debut, which valued Musk's company at $1.77 trillion. SpaceX priced its shares at $135 and jumped 19% to $160 on its June 12 first day. The stock recently traded around $147.
Analysts expect whichever leading AI lab lists first to set valuation benchmarks for the sector. OpenAI confidentially filed in June and is expected to list by early 2027, according to media reports, as its chief rival has signaled a slower path to market.
Market conditions could also make the timing difficult. AI and chip stocks have sold off recently, and Anthropic's listing will test whether investor appetite for the AI trade survives closer scrutiny of its growth projections. A successful deal would nonetheless cap one of the strongest years for US IPOs since 2021.
Valuations and ethical fit are top of mind nowadays, but questions of liability are also crucial for RIA sellers worried about post-sale risks.
Goldman's president and COO is expected to replace David Solomon as CEO as soon as 2027, ending a near-decade at the firm's helm.
A state-by-state Medicaid report card, federal cuts starting in January and a home-equity cap due in 2028 are pushing a program most affluent families ignore into the planning conversation.
New model blends public and private markets as demand for alternatives among wealthy clients accelerates.
Salespeople at the firm often went beyond the matching algorithm to recommend network advisors on its Zoe Wealth platform, according to the regulator.
As AI makes financial information more accessible than ever, Lana Hock explains why human judgment, trust, and empathy remain the qualities clients value most in a financial advisor
Northern Trust’s Ken Lassner shows advisors how to convert volatility into after-tax portfolio gains
