H100 CNX
Data Center GPU
Hopper202280 GB HBM2e350 W
14,592 CUDA cores · 80 GB HBM2e · 350 W
GPU Wiki Series
NVIDIA H-Series
Hopper data center GPU family. Architecture coverage: Hopper.
Quick Facts
- Architecture
- Hopper
- Launch range
- 2022-2024
- Memory types
- HBM2e, HBM3, HBM3e
GPU entries by product stack
Desktop and laptop GPUs are listed in product-stack order.
Data Center GPUs
12 entriesH100 NVL 94 GB
Data Center GPU
Hopper202394 GB HBM3400 W
16,896 CUDA cores · 94 GB HBM3 · 400 W
H100 PCIe 80 GB
Data Center GPU
Hopper202280 GB HBM2e350 W
14,592 CUDA cores · 80 GB HBM2e · 350 W
H100 PCIe 96 GB
Data Center GPU
Hopper202396 GB HBM3700 W
16,896 CUDA cores · 96 GB HBM3 · 700 W
H100 SXM5 64 GB
Data Center GPU
Hopper202364 GB HBM3700 W
16,896 CUDA cores · 64 GB HBM3 · 700 W
H100 SXM5 80 GB
Data Center GPU
Hopper202280 GB HBM3700 W
16,896 CUDA cores · 80 GB HBM3 · 700 W
H100 SXM5 94 GB
Data Center GPU
Hopper202394 GB HBM3700 W
16,896 CUDA cores · 94 GB HBM3 · 700 W
H100 SXM5 96 GB
Data Center GPU
Hopper202396 GB HBM3700 W
16,896 CUDA cores · 96 GB HBM3 · 700 W
H200 NVL
Data Center GPU
Hopper2024141 GB HBM3e600 W
16,896 CUDA cores · 141 GB HBM3e · 600 W
H200 SXM 141 GB
Data Center GPU
Hopper2023141 GB HBM3e700 W
16,896 CUDA cores · 141 GB HBM3e · 700 W
H800 PCIe 80 GB
Data Center GPU
Hopper202380 GB HBM2e350 W
14,592 CUDA cores · 80 GB HBM2e · 350 W
H800 SXM5
Data Center GPU
Hopper202380 GB HBM3700 W
16,896 CUDA cores · 80 GB HBM3 · 700 W
Series Notes
NVIDIA H-Series became a major AI training and inference generation through H100 and H200 class products. Platform, interconnect and memory capacity define the real deployment.