Elon Musk's large language model startup company xAI plans to build a giant data center in Memphis, Tennessee, pending local approval.
The company claims it is acquiring hundreds of thousands of GPUs to train future models, with Nvidia emails yesterday revealing that some of those chip shipments were originally planned for Tesla.
The project is pending approval by the Memphis Shelby County Economic Development Growth Engine (EDGE), Tennessee Valley Authority, and government authorities.
“The good-paying jobs, the cachet of hosting the world’s most powerful supercomputer, and the significant additional revenues for MLGW this project brings will help support our reliability and grid modernization efforts," Doug McGowen, CEO of utility company MLGW, said.
The project would be the “largest multi-billion dollar investment in the city of Memphis’s history," Ted Townsend, president of the Greater Memphis Chamber, said.
Elon Musk, who called the facility a 'Gigafactory of Compute,' said that he aims for the supercomputer to open by the fall of 2025. "My vision is to build the world’s largest and most powerful supercomputer, and I’m willing to put it in Memphis," he said.
It is not known how large the initial phase of the project would be, nor how many millions or billions will be spent in the first phase.
Musk has claimed that xAI will deploy a 100,000 H100 liquid-cooled training cluster within the next few months, and another 300,000 GPU B200 cluster from next summer.
xAI is believed to currently rent around 16,000 Nvidia H100 GPUs from Oracle Cloud. It also uses Amazon Web Services and spare capacity at X/Twitter data centers.
