Back to all GPUs

NVIDIA H100 80GB PCIe Cloud GPU Specs

HexGrid Cloud rents the NVIDIA H100 80GB PCIe at $2.75 per GPU hour, billed per minute — about 8% below the $2.99/hr average across 1 other clouds. The H100 PCIe carries 80GB of HBM2e memory on the Hopper architecture.

80GB VRAMHopperDeploys in ~1 minSecure Cloud

HexGrid Cloud price

$2.75/hr

Billed per minute

VRAM

80GB

HBM2e

FP32

48 TFLOPS

Compute performance

Market average

$2.99/hr

You save 8%

Running the H100 PCIe on HexGrid Cloud

What you get when you deploy with us, beyond the hourly rate.

Live in under a minute

Pick a region, hit deploy, and SSH in. No quota requests, no sales call.

Per-minute billing

You pay for the minutes you use. Stop the instance and billing stops with it.

Ready for training day one

CUDA, PyTorch and the usual drivers are preinstalled, or bring your own image.

No egress charges

Move checkpoints and datasets out without a surprise bandwidth bill.

H100 PCIe pricing across the market

Published on-demand rates from other clouds, so you can see where our H100 PCIe price sits.

Market average $2.99/hr across 1 cloud

8% below market average
  • HexGrid CloudYou're here$2.75/hr
  • RunPod$2.99/hr

Other providers' rates are their published on-demand list prices. They exclude committed-use discounts and can change at any time. Shown for reference only.

About the NVIDIA H100 80GB PCIe

What the H100 PCIe is built for, and where it falls short.

The NVIDIA H100 PCIe is a Hopper Tensor Core GPU designed for high-performance AI and HPC. It adds FP8 acceleration and Transformer Engine technology while supporting MIG, NVLink, confidential computing, and enterprise virtualization.

Best suited for

The H100 PCIe delivers Hopper Tensor Cores, FP8 acceleration, 80GB memory, MIG, and high AI throughput in a PCIe server form factor.

  • LLM inference
  • LLM training
  • Generative AI
  • Model fine-tuning
  • HPC
  • Scientific computing

Strengths

  • FP8 Transformer Engine
  • 80GB high-bandwidth memory
  • MIG support
  • Strong FP64 and AI performance
  • PCIe server compatibility

Limitations

  • Lower bandwidth than H100 SXM
  • Lower NVLink bandwidth than SXM
  • Less memory than H200

NVIDIA H100 80GB PCIe specifications

Full technical specifications for the NVIDIA H100 80GB PCIe.

VendorNVIDIAArchitectureHopperChipGH100CategoryDatacenter GPUReleased2022Form factorPCIe dual-slotInterfacePCIe 5.0 x16

Memory

Memory
80 GB
Type
HBM2e
Bandwidth
2,000 GB/s
Bus width
5120-bit
ECC
Yes

Compute

CUDA cores
14,592
Tensor cores
456
Tensor generation
Gen 4

AI & compute performance

FP64
24 TFLOPS
FP32
48 TFLOPS
TF32 Tensor
400 TFLOPS / 800 TFLOPS sparse
BF16 Tensor
800 TFLOPS / 1,600 TFLOPS sparse
FP16 Tensor
800 TFLOPS / 1,600 TFLOPS sparse
FP8 Tensor
1,600 TFLOPS / 3,200 TFLOPS sparse
INT8 Tensor
1,600 TOPS / 3,200 TOPS sparse

Power & physical

TDP
350 W
Power connector
16-pin
Form factor
PCIe dual-slot
Cooling
Passive
Length
267 mm

Interconnect

Interface
PCIe 5.0 x16
NVLink
Yes
NVLink bandwidth
600 GB/s
Infinity Fabric
No

Virtualization

MIG
Yes
vGPU
Yes
SR-IOV
Yes

Media engines

NVENC engines
0
NVDEC engines
7
AV1 encode
No
AV1 decode
Yes

Software

CUDA
Yes
ROCm
No

Precision support

FP64
Yes
FP32
Yes
TF32
Yes
BF16
Yes
FP16
Yes
FP8
Yes
INT8
Yes

H100 PCIe frequently asked questions

Common questions about NVIDIA H100 80GB PCIe pricing, specifications and workloads on HexGrid Cloud.

How much memory does H100 PCIe have?

The NVIDIA H100 PCIe has 80GB of HBM2e memory with approximately 2 TB/s of memory bandwidth.

Does H100 support FP8?

Yes. Hopper introduces FP8 Tensor Core acceleration and the NVIDIA Transformer Engine.

Does H100 PCIe support MIG?

Yes. The H100 supports up to seven MIG instances.

What is H100 used for?

H100 is designed for LLM training, inference, generative AI, scientific computing, and high-performance computing.

How much does an H100 cost per hour?

H100 cloud rates vary significantly by provider, region, form factor, and GPU count. HexGrid compares current offers.

Other GPUs on HexGrid Cloud

If the H100 PCIe isn't the right fit, these are also available to deploy.

Browse every GPU we rent

Start training on the H100 PCIe today

$2.75 per GPU hour, billed per minute, running in about a minute.

Specification sources