Anthropic blinda 80.000 millones de dólares en capacidad de cómputo antes de su salida a Bolsa

Anthropic, the creator of Claude, has secured $8 billion in new deals with Lambda and Nscale to ensure access to computing chips and data centers ahead of its planned stock market debut. Anthropic has committed to spending $800 million on new AI infrastructure contracts in recent weeks, a move aimed at securing access to chips and data centers before its IPO.

The latest agreement closed this week with Lambda, a cloud company backed by Nvidia, for $350 million. Nvidia acts as a guarantor for the project after securing a lease agreement with Hut 8, a company building a data center in Texas, according to The Wall Street Journal. Nvidia thus guarantees demand for its AI processors in a new mega-data center, with Lambda managing its operational execution.

Last week, Anthropic agreed to spend $450 million renting computational power in Nscale's data center in West Virginia, which will also utilize Nvidia's chips, under a similar arrangement. Anthropic's aggressive push reflects the need to shore up computational capacity amid the race among major labs for resources in data centers, as demand outpaces available supply.

The company, which previously struggled with handling spikes in usage due to a lack of infrastructure, has intensified efforts to secure supply and mitigate one of the risks that investors will scrutinize ahead of its IPO. Wall Street anticipates Anthropic's stock market debut this summer, a potentially record-breaking operation surpassing SpaceX's IPO earlier this year.

Investors are considering an IPO valuation of up to $100 billion, potentially reaching $2 trillion in valuation. In May, Anthropic's leader, Dario Amodei, signed a $450 million mega-deal with SpaceX to rent computational power in its AI data centers. Earlier this month, Anthropic secured a multi-year agreement for over $100 billion worth of computing power and Nvidia chips with Amazon as part of a circular agreement where Amazon invests $25 billion in the startup.

Anthropic also signed a five-year agreement with Google for another 5 GW of computational capacity in April, which includes the use of the tech giant's AI chips and makes Google a shareholder in Anthropic. These alliances follow a circular pattern under scrutiny by investors, as tech giants act as clients, suppliers, and investors for these AI companies.