From NVIDIA

Blackwell SXM module designed as a drop-in for HGX H100 baseboards. Lower-TDP sibling of B200.

2026-05-01 · verified 1mo ago

Blackwell SXM module. Dual-die GPU with HBM3e, fifth-gen NVLink. Building block of HGX B200, DGX B200, and GB200.

2026-05-01 · verified 1mo ago

Air-cooled DGX. Eight Blackwell GPUs on an HGX baseboard, dual Xeon Platinum, BlueField-3, ConnectX-7.

2026-05-01 · verified 1mo ago

Eight-way H200 server with HBM3e, FP8 Tensor Cores, dual Xeon Platinum, NVSwitch fabric.

2026-05-01 · verified 1mo ago

GB10 Grace Blackwell mini-workstation: unified memory, ConnectX-7, desk-sized chassis.

2026-05-01 · verified 1mo ago

NVIDIA's in-house Hopper supercomputer — exa-class FP8 across hundreds of DGX H100 nodes.

2026-05-01 · verified 1mo ago

One Grace CPU coherently linked to two Blackwell GPUs over NVLink-C2C. Unit cell of the NVL72.

2026-05-01 · verified 1mo ago

Liquid-cooled rack — Blackwell GPUs and Grace CPUs as one NVLink domain, exaflop-class FP4.

2026-05-01 · verified 1mo ago

Blackwell mid-tier: 16 GB GDDR7 on a 256-bit bus, 8,960 CUDA cores, 300 W TGP.

2026-05-01 · verified 1mo ago

Blackwell second-tier: 16 GB GDDR7 at 30 Gbps, 10,752 CUDA cores, 360 W TGP, 2-slot reference cooler.

2026-05-01 · verified 1mo ago

Blackwell flagship: 32 GB GDDR7, 21,760 CUDA cores, 575 W TGP, 2-slot Founders form factor.

2026-05-01 · verified 1mo ago

Grace Arm CPU and Hopper GPU on one module, joined by NVLink-C2C, with HBM3e on the GPU side.

2026-05-01 · verified 1mo ago

Hopper-generation SXM accelerator: HBM3, fourth-gen Tensor Cores, the workhorse of the LLM training era.

2026-05-01 · verified 1mo ago

Hopper-generation SXM accelerator with HBM3e, drop-in for HGX H100 baseboards.

2026-05-01 · verified 1mo ago

Eight-GPU Blackwell baseboard. HBM3e, NVLink 5 fabric, foundation of OEM Blackwell servers.

2026-05-01 · verified 1mo ago

Blackwell-generation edge module: native FP4, LPDDR5X, Arm Neoverse-V3AE, configurable power envelope.

2026-05-01 · verified 3w ago

Workstation entry: 32 GB GDDR7 ECC, 10,496 CUDA cores, 200 W, dual-slot.

2026-05-01 · verified 1mo ago

Workstation tier: 48 GB or 72 GB GDDR7 ECC, 14,080 CUDA cores, 300 W, dual-slot.

2026-05-01 · verified 1mo ago

Workstation flagship: 96 GB GDDR7 ECC, 24,064 CUDA cores, 600 W max, dual-slot.

2026-05-01 · verified 1mo ago

24 GB. Still fast. Still honest.

2026-04-24 · verified 1mo ago