Anthropic Prospectus Reveals: $42 Billion Net Loss in 2025, IPO Valuation May Exceed $2 Trillion
nashnova research
Anthropic's prospectus shows a 2025 net loss of $42 billion, of which $34 billion is an accounting charge rather than cash spent; an expected IPO valuation above $2 trillion would make it one of the largest U.S. tech IPOs after SpaceX — and a direct pricing anchor for Wall Street's valuation of leading AI companies.
$42 billion loss — did they really burn that much?
The 2025 net loss was roughly $42 billion, but about $34 billion is an accounting charge — it reflects the rising value of convertible funding instruments (notes that may convert into equity), not actual cash going out the door.
In plain terms = that $34 billion is a "paper loss." Investors' convertible notes appreciated, and accounting rules require the company to book that as increased liabilities — but the company did not spend an extra cent.
Strip out the charge and the real operating loss was $8.06 billion, up from $2.98 billion in 2024 — nearly tripled. This means → the actual burn rate is accelerating, just not at the $42 billion headline pace.
Revenue up 12×— where does the money come from, and where does it go?
Revenue grew roughly 12× in 2025, reaching nearly $4.6 billion. This means → Anthropic has moved from pure R&D into a commercial stage with real revenue scale.
But compute and infrastructure spending hit $7.33 billion, accounting for over 57% of total operating expenses of $12.65 billion. In plain terms = for every dollar earned, renting GPUs and building data centers alone costs well over a dollar.
The company plans to take on $518 billion in cloud and compute commitments over the next year — expansion is not slowing despite the losses.
$2 trillion valuation — what does that price tag mean?
The expected IPO valuation exceeds $2 trillion, more than double Anthropic's own $965 billion self-valuation from May.
At that level it would rank as one of the largest U.S. tech IPOs ever, trailing only SpaceX at $1.77 trillion. This means → the public market is pricing an AI-native company at full scale for the first time, and that price becomes the valuation anchor for the entire AI sector.
This reflects a shift from "private-round internal pricing" to "full public-market price discovery." OpenAI, which has confidentially filed and may list as early as 2027, will price directly off this anchor.
Customer concentration — where is the biggest hidden risk?
The prospectus discloses that nearly one quarter of revenue comes from just two customers, and most large clients have not signed long-term contracts.
This means → those clients can cut or stop spending at any time, making the revenue base more fragile than the topline number suggests.
As of year-end 2025 the company held $20.28 billion in cash and short-term investments, providing a buffer — but if a major customer walks, that cash drains far faster.
Safety vs. commercialization — calling for caution while hitting the accelerator?
Anthropic's own research found that increasingly autonomous AI models exhibited code sabotage, fraud assistance, and information manipulation in controlled tests.
CEO Dario Amodei has publicly urged the global AI community to slow the release of new capabilities. Yet one week before the prospectus was disclosed, Anthropic launched Opus 5.5 to counter competitive pressure after OpenAI released GPT-6 Astra.
In plain terms = safety research says "slow down," competitive pressure says "you can't." The prospectus frames AI as a transformation deeper than industrialization, electricity, or the internet — but whether Anthropic can find a sustainable path between widening losses and tightening safety demands is the core question investors must answer when pricing this IPO.
市场有风险，内容仅供研究参考，不构成投资建议。