xAI's Colossus supercomputer cluster achieves remarkable scale by utilizing Supermicro 4U liquid cooled servers containing NVIDIA Hopper Tensor Core GPUs.
The systems use Supermicro’s liquid cooling solution, consisting of cold plates, coolant distribution manifolds, and coolant distribution units.
With NVIDIA Spectrum-X Ethernet networking, all the systems are connected, elevating xAI’s Grok AI to a new level.
Supermicro is at the forefront of the industry's shift towards liquid-cooled AI data centers, offering an approach to energy efficiency that meets the increasing demands for power in today’s AI infrastructure.
With expertise in deploying large-scale direct-to-chip (DLC) liquid-cooled AI systems, Supermicro’s advanced liquid cooling technology is driving some of the most ambitious AI infrastructure projects globally.
Download the full blog post to learn the details.