- Anthropic’s IPO prospectus shows a rapidly expanding AI business with billions in revenue but equally large infrastructure requirements, as the Claude developer prepares for a potential public listing at a valuation above $2 trillion.
Anthropic generated $4.59 billion in revenue in 2025, a roughly twelvefold increase from the previous year, while reporting an $8.06 billion operating loss, according to the IPO prospectus reviewed by Reuters.
The company also reported a net loss of approximately $41.97 billion for the year. Reuters said the figure was heavily affected by an accounting charge of roughly $34 billion related to the revaluation of financing instruments that could convert into equity. This accounting charge makes the operating-loss figure a more useful indicator of the underlying cost of running the business, although it does not make Anthropic’s infrastructure requirements small.
The company spent approximately $7.33 billion on computing and infrastructure in 2025, nearly three times the prior year's amount, according to the prospectus. Anthropic ended the year with approximately $20.28 billion in cash, cash equivalents, and short-term investments, according to reporting based on the document.
The prospectus also outlines approximately $518 billion in future cloud, computing and infrastructure obligations, underscoring the amount of capital required to develop and operate increasingly capable AI systems.
Anthropic's Revenue Is Growing Fast, But So Are Its Infrastructure Costs
Anthropic's growth has accelerated sharply as businesses adopt Claude and related AI products. The company's reported 2025 revenue of $4.59 billion compares with roughly $400 million in 2024. The company later said its annualized revenue had climbed much further during 2026 as demand for Claude increased.
Its trajectory is central to the potential IPO valuation. Reuters reported that Anthropic could seek a valuation of more than $2 trillion, more than double the company's estimated $965 billion valuation from its May financing round.
The scale of Anthropic's computing commitments reflects the economics of frontier AI. Training and operating advanced models require large amounts of computing capacity, and demand increases as models are deployed across more applications and customers.
Anthropic has also been building relationships across the broader cloud and computing industry to secure that capacity.
The company said in its Responsible Scaling Policy that its safeguards are intended to scale alongside the capabilities and risks of its models. Its latest framework includes measures covering security, safeguards, alignment and policy. The infrastructure requirement is particularly relevant because Anthropic's revenue is not evenly distributed across its customer base.
Reuters reported that two customers each accounted for about 12% of 2025 revenue, meaning the two largest customers represented roughly 24% of sales combined. The prospectus also warns that some major customers do not have long-term contracts, leaving revenue exposed to changes in spending by large clients.
That customer concentration is one of the financial risks public-market investors will have to assess alongside Anthropic's rapid growth. It also provides a useful comparison with other frontier AI companies. OpenAI's revenue run rate has surpassed $40 billion, illustrating the scale at which leading AI companies are beginning to monetize enterprise and consumer demand.
IPO Filing Highlights AI Safety and Public-Market Risks
Anthropic's prospectus devotes substantial space to risks associated with both the business and the technology itself. The Financial Times reported that roughly 80 pages of the 261-page prospectus were devoted to risk factors, compared with about 48 pages describing Anthropic's business.
Among the risks described are potential behaviors from highly capable AI models that could be difficult to predict or control. Reuters reported that the prospectus warns models could potentially display self-preserving behavior, resist shutdown attempts, conceal or manipulate information, or exhibit behavior resembling blackmail.
Anthropic's own safety framework provides additional context for those disclosures. The company says its Responsible Scaling Policy is designed to address risks that could emerge as AI capabilities advance and includes escalating safeguards for more capable systems.
The prospectus also presents a more conventional set of financial risks, including customer concentration, substantial infrastructure commitments and the possibility that computing expenses could continue rising faster than revenue.
That combination is central to the Anthropic IPO story. The company's revenue growth demonstrates significant commercial demand for Claude, but its operating results show that frontier AI remains an extremely expensive business to scale.
The potential public listing would give investors one of their clearest opportunities to evaluate those economics directly. The timing remains uncertain. Reuters reported that Anthropic could pursue the IPO after the November 2026 US midterm elections, subject to market conditions and the company's plans.
For investors, the most important figures currently disclosed are straightforward: $4.59 billion in 2025 revenue, an $8.06 billion operating loss, a $41.97 billion net loss, $7.33 billion in computing and infrastructure spending, $20.28 billion in year-end liquidity, and approximately $518 billion in future infrastructure obligations.
Those numbers provide a clearer picture of both the opportunity and the financial demands behind Anthropic's attempt to become one of the world's largest AI companies.