NVIDIA and xAI Launch World's Largest AI Supercomputer Colossus with 100,000 GPUs in 122 Days, Achieving 95% Data Throughput

Oct 28, 2024, 03:37 PM

NVIDIA has announced the completion of the world's largest AI supercomputer, Colossus, built in just 122 days by xAI and featuring 100,000 NVIDIA Hopper GPUs. Located in Memphis, Tennessee, Colossus utilizes NVIDIA's Spectrum-X Ethernet networking platform, achieving 95% data throughput and eliminating latency issues. Elon Musk, a key figure behind xAI, praised the project, stating that Colossus is the most powerful training system globally. Plans are underway to expand the supercomputer's capacity to 200,000 GPUs, further enhancing its capabilities in artificial intelligence research and development.