Anthropic launched an aggressive effort to secure more computing power after executives realized late last year that the rapidly growing company could run short of the capacity required for artificial intelligence development and products.
Chief executive Dario Amodei described the challenge as a “cone of uncertainty” at The New York Times’s DealBook Summit in December, saying the wide range of possible demand made it difficult to avoid mistakes in planning.
In the months that followed, the San Francisco-based AI company signed dozens of contracts with former cryptocurrency miners, data center start-ups, chipmakers and major technology companies including Amazon, Google, Microsoft and SpaceX, according to people familiar with the company, investors and industry experts.
Anthropic also hired data center specialists and purchased equipment intended to accelerate construction. Its representatives met with developers, examined the possibility of converting former coal plants and factories into data centers, and considered arrangements involving small amounts of computing capacity that could collectively meet its needs.
“Anthropic is gobbling up a lot of the capacity,” said Erik Bernhardsson, chief executive of computing provider Modal.
The effort is central to Anthropic’s ambitions in artificial intelligence. Computing power is needed both to develop advanced models and to operate products such as the Claude chatbot for businesses and consumers. The company devoted more than 10 pages of its public offering prospectus to explaining the importance and risks of securing that resource as it pursues what could become the largest initial public offering of its kind this month.
Sources: The New York Times Technology.