Elon Musk's xAI Receives Approval For 150MW, Allowing 100,000 GPUs To Operate Simultaneously
Elon Musk's xAI data center, dubbed the "Gigafactory of Compute" or "Colossus," represents a significant milestone in artificial intelligence and supercomputing. Located in Memphis, Tennessee, this facility has recently been approved to receive 150 megawatts (MW) of power from the Tennessee Valley Authority (TVA), enabling all 100,000 Nvidia H100 GPUs to operate concurrently.
Technological Advancements
The xAI Colossus is built on cutting-edge technology, utilizing Nvidia's H100 GPUs, which offer up to nine times the speed of the previous A100 models. Each H100 GPU delivers up to 2,000 teraflops of performance, making the initial setup of 100,000 GPUs a formidable force in the realm of supercomputing. The facility plans to expand further by adding 50,000 more H100 GPUs and 50,000 new H200 GPUs, potentially achieving a theoretical performance of approximately 497.9 exaflops.
Power and Infrastructure
The approval for 150 MW of power marks a significant increase from the initial 8 MW supply, raising concerns about the impact on local power reliability and prices. The facility's energy demand is equivalent to the needs of approximately 100,000 homes, and the power is partially supplemented by onsite gas turbines.
