Claude maker Anthropic reportedly plans to bet big on cloud and computing in the coming years, despite recording significant losses last year.
Citing Anthropic's confidential IPO prospectus, Reuters reported Monday that the company's revenue grew 12-fold last year to nearly $4.6 billion, and that nearly a quarter of the revenue came from two customers.
The company posted a net loss of $42 billion, including more than $8 billion in losses on an operating basis. It spent $7.33 billion on computing and infrastructure last year, a threefold increase from the previous year. That also accounted for over half of the company's $12.65 billion in total operating expenses.
Despite the losses, the company said it plans to spend $518 billion on cloud, computing and infrastructure obligations in the coming years, according to the report.
Anthropic confidentially filed for an IPO in June and is reportedly seeking to raise $100 billion at a valuation of around $2 trillion. Such a valuation would place its IPO among the largest tech debuts, following SpaceX's recent public listing that valued the Aerospace company at $1.77 trillion.
Its rival, OpenAI, has also confidentially filed for an IPO in June and is reportedly seeking to list by early 2027.
Safety risks
Anthropic also plans to caution investors in the prospectus about risks associated with training frontier AI models, as major AI firms have called for slowing the pace of development.
According to a separate Reuters report, Anthropic's prospectus warns that AI could pose "catastrophic or existential risks to humanity," particularly if advanced AI models exhibit "self-preserving behaviors." Such behaviors could include attempts to resist shutdown, conceal or manipulate information, or engage in conduct resembling blackmail. Anthropic devoted roughly 80 pages of the prospectus's 261-page main body to outlining risk factors, per the report.
Safety has become a prominent part of discussions around AI training. On Monday, OpenAI released safety guidelines, stating that structured security documentation should be required before continuing any frontier reinforcement learning training run.
Anthropic CEO Dario Amodei and OpenAI CEO Sam Altman also spoke last week at the UN General Assembly, calling for greater coordination in responsible AI development.
"We will slow down as much as necessary in order to make sure that every successive AI technology that we release is actually safe," Amodei said.
Amodei further urged countries to create universal standards for safety tests and systems to alert one another to AI incidents, particularly those involving national security.
Notably, Australian Prime Minister Anthony Albanese revealed last week that an OpenAI agent hacked the country's Medicare service portal in June, in what may be the first known instance of an AI agent breaking into a government system. Albanese said no personal information was leaked, but added that "this situation is obviously unacceptable."