Hopper

NVIDIA H100 PCIe

Hopper performance in a standard PCIe form factor

VRAM

80 GB

Bandwidth

2.0 TB/s

FP16

756.5 TFLOPS

TDP

350W

Get H100 PCIe Pricing View Comparisons
NVIDIA H100 PCIe

Technical Specifications

VRAM 80 GB HBM3
Memory Bandwidth 2.0 TB/s
FP16 Performance 756.5 TFLOPS
BF16 Performance 756.5 TFLOPS
FP32 Performance 51.2 TFLOPS
INT8 Performance 1,513 TOPS
TDP 350W
Form Factor PCIe Gen5 Dual-Slot
Interconnect NVLink Bridge (600 GB/s, 2-GPU pairs)
PCIe Interface PCIe Gen5 x16
Max GPUs per Server 4-8 (chassis dependent)

Prices vary with supply and import costs. Contact for current India pricing.

Best For

AI inference at scale with lower power budgets
Fine-tuning and training of mid-size models (7B-30B parameters)
Drop-in upgrade path from A100 PCIe in existing servers
Multi-tenant GPU servers for cloud and managed hosting

Not Ideal For

  • Maximum-throughput multi-GPU training (SXM version has 900 GB/s NVLink vs 600 GB/s)
  • Workloads requiring more than 2-GPU NVLink connectivity

Overview

The NVIDIA H100 PCIe brings Hopper-generation performance to standard PCIe server chassis. While it trades some peak throughput compared to the SXM variant (756 vs 989 FP16 TFLOPS) and has lower memory bandwidth (2.0 vs 3.35 TB/s), it offers a significantly lower TDP of 350W and compatibility with a wider range of servers.

The PCIe form factor supports NVLink Bridge connections between pairs of GPUs, making 2-GPU configurations highly efficient for inference and fine-tuning. For teams upgrading from A100 PCIe deployments, the H100 PCIe is a direct drop-in replacement in most chassis.

This is a strong choice for inference deployments, managed GPU cloud platforms, and organizations that need H100-class performance without the complexity and cost of a full HGX baseboard setup.

Compare

NVIDIA H100 vs A100

Compare NVIDIA H100 SXM and A100 80 GB GPUs for AI training and inference workloads. Detailed specs, performance, and value analysis for Indian enterprises.

View comparison

NVIDIA H100 vs L40S

Compare NVIDIA H100 and L40S GPUs for AI inference, training, and visual compute. Specs, performance, and pricing analysis for Indian data centres.

View comparison

Use Case Guides

Best GPU for LLM Inference in India

Large Language Model (LLM) Inference / Serving

Read guide

Best GPU for LLM Training in India

Large Language Model (LLM) Training

Read guide

Get NVIDIA H100 PCIe pricing for your setup

Tell us your workload and cluster size. We'll quote the complete solution including servers, networking, and colocation.

Get a Quote WhatsApp Us