The HGX H200 platform brings together eight H200 GPUs, totaling 1.13TB of blazing-fast HBM3e memory. Purpose-built for LLM training, HPC, and enterprise-grade AI, this platform delivers peak transformer performance and exceptional data throughput.

• Memory: 1.13TB HBM3e (141GB per GPU)

• Connectivity: NVLink for ultra-fast GPU communication

• Form Factor: SXM baseboard

• Price: $300,000 – $350,000 (allocation-based)

• Availability: Limited