AWS has launched one of the largest operational AI data centers in the world, with its supercomputer complex Project Rainier in Indiana.
The 1,200-acre site near Lake Michigan will be used to train Anthropic’s AI model Claude, using nearly half a million of AWS’ in-house Trainium2 chips and with plans to scale up to more than a million by the end of this year.
Unlike general-purpose chips, AWS said Trainium2 is tailored to process “the enormous amounts of data” needed to train AI models in complex tasks.
With Project Rainier, AWS has already built Trainium2 infrastructure that’s 70% larger than any other AI computing platform in AWS history and offers more than five times the compute power Anthropic used to train its previous AI models.
Plans for the project were first announced last year, when AWS said it would build an AI compute cluster spread across multiple data centers in the U.S.
The $11 billion Rainier project represents the first stage of this initiative, and a major milestone in AWS’ commitment to rapidly scaling up its AI infrastructure.
“Project Rainier is one of AWS’ most ambitious undertakings to date,” said Ron Diamant, head architect of Trainium. “It’s a massive, one-of-its-kind infrastructure project that will usher in the next generation of artificial intelligence models.”
While it has not disclosed the full scope of the project, AWS will reportedly construct an additional 23 buildings in the future, set to increase the site’s data center capacity to more than 2.2 gigawatts.
The news also comes as tech giants race to pull out ahead in the AI chip landscape.
Last week, Anthropic said it would use 1 million of Google’s custom chips, in a partnership Anthropic said was worth “tens of billions” of dollars.
On Oct. 15, Nvidia, Microsoft, xAI and BlackRock agreed on a $40 billion deal to buy Aligned Data Centers, while in September Nvidia invested $100 billion in Anthropic rival and ChatGPT creator OpenAI.
