top of page
B200 Server Platform – 512 Units Available
AI Supercomputing in a 10U Form Factor
Why the B200 Platform?
Scalable AI Power: Handle trillion-parameter models and multi-node training.
Speed & Redundancy: Built-in Infiniband ensures maximum performance at scale.
Secure Access: EUA required—request to begin procurement.
Built for generative AI, LLM training, and hyperscale applications, the B200 Server is a next-gen compute monster. Featuring NVIDIA HGX B200 with 8× Umbriel GPUs and 1.44TB of HBM3e, it’s the ultimate stack for serious innovators.
CPU: 2× Intel Xeon Platinum 8562Y+ (64 cores)
GPU: 8× B200 Umbriel (180GB HBM3e each)
RAM: 2TB DDR5 ECC
Storage: 2× 1.92TB NVMe SSDs
Networking: 8× ConnectX-7 NICs (400Gbps IB/Ethernet)
Form Factor: 10U Rackmount
bottom of page