Anthropic is preparing to offer its shares to the public at a target valuation of up to two trillion dollars, with plans to spend approximately $518 billion on computing, chips, and infrastructure. The offering prospectus reveals significant losses, revenue concentration, and extensive warnings about the risks of advanced artificial intelligence models.
Anthropic is approaching a public offering at a target valuation of up to two trillion dollars, according to Reuters, based on the offering prospectus. If achieved, this valuation would be more than double the company’s valuation of $965 billion four months ago and could place the offering in competition for the title of the largest in history, surpassing the figure associated with SpaceX.
A Massive Financial Bet on Infrastructure
According to the prospectus, the company plans to spend approximately $518 billion over the coming years on cloud-computing, chip, and infrastructure commitments. This reflects the dependence of its business model on providing extensive computing capacity to train and operate Claude models, rather than viewing revenue growth alone as sufficient evidence that profitability is approaching.
Anthropic’s revenue increased approximately twelvefold during 2025 to approach $4.6 billion, but it recorded a loss of $42 billion that year, while losses from commercial operations alone exceeded $8 billion. The company relies on revenue from the use of its models based on the number of processed tokens, in addition to customer subscriptions to Claude models.
Revenue Concentration Increases Offering Risks
The financial concern is not related only to the scale of the losses. The Financial Times reported that approximately one-quarter of Anthropic’s revenue during 2025 came from two customers, meaning that a significant portion of the company’s performance depends on a limited number of customers. This concentration gives investors an additional reason to examine the sustainability of demand before accepting a valuation of two trillion dollars.
What Does the Offering Prospectus Reveal About the Risks?
Anthropic devoted approximately 80 of the 261 pages in its offering prospectus to discussing risks associated with advanced artificial intelligence models. It warned that developing more advanced models, platforms, and applications and expanding their use could lead to risks it described as “catastrophic or existential” for humanity.
The prospectus presents the results of tests that the company said it conducted on its models, including behaviors such as attempting to conceal or manipulate information, blackmailing users in certain scenarios, and resisting attempts to shut down the models in ways associated with self-preservation behaviors. Reports also quoted Evan Huberger, an artificial intelligence safety researcher at Anthropic, as estimating earlier in September that the likelihood of artificial intelligence causing human deaths during the next decade exceeds 10 percent.
Why Does This News Matter?
The potential offering presents investors with a clear paradox: extremely rapid revenue growth is accompanied by substantial infrastructure spending and massive losses, as well as dependence on a limited number of customers, while the company itself acknowledges fundamental risks in the technology that forms the basis of its business. The available information does not determine whether the target valuation reflects a sustainable ability to generate returns or optimistic expectations regarding the expansion of the artificial intelligence market.