Inferbase
Beta
NVIDIA H100 PCIe
Datacenter
The H100 PCIe brings Hopper architecture to standard server form factors with 80GB HBM3 memory. Ideal for enterprises requiring high AI performance without SXM infrastructure.
VRAM
80 GB
Memory
HBM3
Bandwidth
2000 GB/s
TDP
350W
Suitable Workloads
Large Language Models
Training and inference for models like GPT-4, Llama 70B+
Deep Learning Training
High-performance training for neural networks
Distributed Training
Multi-node training with fast interconnects
High-Throughput Inference
Optimized for batched inference workloads
Key Highlights
- Built on Hopper architecture
- 456 AI accelerator cores for tensor operations
- 14,592 compute cores for parallel processing
- PCIe form factor
Memory
VRAM80 GB
Memory TypeHBM3
Memory Bandwidth2000 GB/s
Compute Performance
FP32 (Single Precision)756 TFLOPS
FP16 (Half Precision)1513 TFLOPS
BF161513 TFLOPS
INT83026 TOPS
Architecture
ArchitectureHopper
Compute Cores14,592
AI Accelerators456
Release Year2022
Power & Physical
TDP350 W
Max Power350 W
Form FactorPCIe
PCIe GenerationGen5 x16
Interconnect
Multi-GPU SupportYes
Interconnect Bandwidth600 GB/s
Pricing
MSRP$25,000
Notes
PCIe form factor
Models That May Fit on NVIDIA H100 PCIe
Estimates based on INT8 quantization. Actual fit depends on framework and batch size.
Related Tools
Added Jan 25, 2026
Last updated: Jan 25, 2026
Build Your AI Stack with Confidence
Explore models, compare pricing and benchmarks, and right-size your infrastructure — all in one place.