Anthropic has filed for an initial public offering that could value the artificial intelligence company at more than $2 trillion, according to its 261-page prospectus. The proposed valuation, if achieved, would more than double the $965 billion estimate cited in May and place the Claude developer on course to surpass SpaceX’s reported record for the highest-valued IPO.
The filing presents an AI company pursuing expansion on a scale rarely seen in public markets. Anthropic reported $4.59 billion in 2025 revenue, up 1,088% from a year earlier, while forecasting $518 billion in spending over the next year on cloud computing, compute capacity and infrastructure. That planned outlay is about 113 times its 2025 revenue and roughly 26 times the $20.28 billion it held in cash, cash equivalents and short-term investments at the end of December.
A listing timetable remains uncertain. Prospectus-related reporting cited the possibility that the offering could move beyond the U.S. midterm elections in November. OpenAI, Anthropic’s largest private-market rival, filed for an IPO in June but has reportedly pushed its own market debut to early 2027.
Revenue growth comes with escalating infrastructure costs
Anthropic’s prospectus shows that commercial demand for its models expanded rapidly during 2025, even as the costs of serving and improving those systems climbed faster than its revenue base.
The company’s operating loss exceeded $8 billion in 2025, compared with a $2.98 billion operating loss in 2024, according to the filing. Its GAAP net loss widened to $41.97 billion from $8.31 billion. Anthropic attributed approximately $34 billion of the latest loss to accounting effects related to increases in the value of financing instruments.
Compute and infrastructure costs reached $7.33 billion in 2025, tripling from the previous year and accounting for 58% of total operating expenses of $12.65 billion. The figures show how intensely large AI developers depend on chips, data-center capacity and long-term cloud commitments, even after reaching multibillion-dollar annual revenue.
Anthropic has raised about $122.654 billion through roughly 10 funding rounds, the prospectus states. Public markets would offer the company another route to finance the data-center commitments required to compete with OpenAI and other frontier-model developers.
The filing also disclosed customer concentration. Anthropic’s two largest direct customers each provided 12% of 2025 annual revenue, or 24% combined. Other major customers do not hold long-term contracts, according to the document, leaving a substantial portion of revenue exposed to renewal decisions and shifting corporate AI budgets.
Anthropic released its Opus 5.5 model on Sept. 22, the filing said, amid competition with OpenAI’s GPT-6 Astra. Product releases have become closely tied to spending decisions: more capable models generally require larger training runs and greater inference capacity, the computing used when customers submit prompts after a model has been trained.
Safety risks occupy a large share of the filing
About 80 pages of Anthropic’s prospectus address risks associated with advanced AI systems, including scenarios involving severe harm to humans. The unusually extensive discussion places safety questions alongside conventional IPO risks such as losses, customer concentration, competition and capital requirements.
The filing says models could exhibit forms of “self-preservation” behavior, including efforts to resist shutdown, conceal or manipulate information, or engage in conduct comparable to blackmail. It also warns that a model could become aware that it is being tested in a safety evaluation, potentially limiting the company’s ability to measure risks accurately.
Anthropic did not disclose a specific amount allocated to AI safety research. The prospectus instead frames safety as a central operational challenge for a company building increasingly capable models while attempting to commercialize them at scale.
Those disclosures could receive particular scrutiny from prospective shareholders because Anthropic’s governance model gives ordinary public shareholders limited influence over key decisions. The company plans to remain a Delaware public benefit corporation after its IPO, a structure that formally allows directors to consider specified public benefits alongside shareholder returns.
Founders retain decisive voting influence
Anthropic plans a five-class share structure that would concentrate voting power with its founders and other designated holders after the IPO. Class A common stock, the shares expected to be broadly available to public buyers, would carry one vote per share but would not control several major corporate decisions.
A proposed Founder LLC would allow the founders to determine the issuance of a single Class F share carrying 50.1% of voting power on specified matters. Those matters include electing certain directors and other issues submitted to shareholders, according to the prospectus.
The company identified seven co-founders: Dario Amodei, Daniela Amodei, Tom Brown, Jack Clark, Jared Kaplan, Sam McCandlish and Chris Olah. The filing describes conditions under which a founder could be removed from the Founder LLC, including departure from the company, death, selling an excessive number of shares or removal for cause.
Anthropic’s board would have seven seats. Daniela Amodei, Dario Amodei and an additional director yet to be named would occupy three seats, while the remaining four would be elected by Long-Term Benefit Trust, or LTBT, through Class T shares. Class A shareholders would not vote directly for those four directors.
The prospectus also says the founders intend to direct 80% of their personally held shares toward charitable purposes. Executive-pay disclosures show Dario Amodei, Anthropic’s chief executive officer, received nearly $18 million in 2025, primarily through stock and option awards. Daniela Amodei, the company’s president, received $16.4 million and was the second-highest-paid executive.
AI capital spending may reshape risk appetite, but crypto links remain unproven
Anthropic’s proposed spending plan illustrates the pressure that AI infrastructure demand is placing on private financing, public equities, cloud providers and semiconductor supply chains. A $518 billion commitment would require continued access to external capital and large counterparties willing to provide computing capacity over extended periods.
That does not provide evidence that an Anthropic IPO will directly pull liquidity from cryptocurrency markets or dictate digital-asset prices. Crypto prices are influenced by many factors, including leverage, macroeconomic conditions, stablecoin flows, token-specific supply schedules and changes in regulation. The prospectus contains no projection linking Anthropic’s capital needs to token-market performance.
The document also does not support claims that tokens associated with decentralized physical infrastructure networks, often called DePIN, would serve as a safe shelter from market volatility. These networks seek to use token incentives to coordinate hardware such as graphics processors, storage devices and wireless equipment, but their tokens can remain highly sensitive to broader risk sentiment and project-specific execution risks.
Anthropic’s filing instead gives public-market participants a detailed view of the trade-off defining the frontier AI race: exceptional revenue growth and demand for advanced models alongside capital requirements, governance restrictions and safety risks that remain unusually difficult to quantify.
For deeper context on AI, markets, and finance, explore our guide to digital assets and why they matter today.
Disclaimer: The content on this page is provided for general informational purposes only and does not represent the views or financial advice of Toobit. We make no guarantees regarding the accuracy or completeness of this information and shall not be held liable for any errors, omissions, or outcomes resulting from its use. Investing in digital assets involves risk; users should independently evaluate their financial situation and the risks involved. For further details, please consult our Terms of Service and Risk Disclosure.
