Skip to content

NVIDIA H100 80GB (SXM): cloud rental cost

Renting one NVIDIA H100 80GB (SXM) around the clock costs from $1,964 a month (Runpod Community Cloud, on demand) to $71,774 (Microsoft Azure, on demand), across 5 providers with a published price. The listed Microsoft Azure and AWS EC2 instances have 8 GPUs, so a single GPU there is billed as 8.

A full eight-GPU setup on demand ranges from $15,710 a month on Runpod Community Cloud to $71,774 on Microsoft Azure. Spot capacity on Microsoft Azure is 82% below its on-demand price, as a snapshot on the date checked. Google Cloud lists this GPU, but no price has been verified yet.

Prices on this page last verified on Sources

1 GPU, 24 hours a day

One GPU serving a model around the clock. Monthly cost, cheapest on-demand provider first.

NVIDIA H100 80GB (SXM), 1 GPU, 24 hours a day: monthly cost by provider
ProviderGPUs billedOn demandSpot1-year reserved
Runpod Community Cloud1$1,964No self-serve price published for this option.No self-serve price published for this option.
Runpod Secure Cloud1$2,913No self-serve price published for this option.No self-serve price published for this option.
Lambda1$3,570No self-serve price published for this option.No self-serve price published for this option.
AWS EC28$40,179Price not verified yet.Price not verified yet.
Microsoft Azure8$71,774$13,264$45,935billed every hour
Google CloudPrice not verified yet.Price not verified yet.Price not verified yet.

8 GPUs, 24 hours a day

A full eight-GPU server running around the clock. Monthly cost, cheapest on-demand provider first.

NVIDIA H100 80GB (SXM), 8 GPUs, 24 hours a day: monthly cost by provider
ProviderGPUs billedOn demandSpot1-year reserved
Runpod Community Cloud8$15,710No self-serve price published for this option.No self-serve price published for this option.
Runpod Secure Cloud8$23,302No self-serve price published for this option.No self-serve price published for this option.
Lambda8$26,806No self-serve price published for this option.No self-serve price published for this option.
AWS EC28$40,179Price not verified yet.Price not verified yet.
Microsoft Azure8$71,774$13,264$45,935billed every hour
Google CloudPrice not verified yet.Price not verified yet.Price not verified yet.

1 GPU, 8 hours a day

Development or test work during a working day, switched off overnight. Monthly cost, cheapest on-demand provider first.

NVIDIA H100 80GB (SXM), 1 GPU, 8 hours a day: monthly cost by provider
ProviderGPUs billedOn demandSpot1-year reserved
Runpod Community Cloud1$654.57No self-serve price published for this option.No self-serve price published for this option.
Runpod Secure Cloud1$970.90No self-serve price published for this option.No self-serve price published for this option.
Lambda1$1,190No self-serve price published for this option.No self-serve price published for this option.
AWS EC28$13,393Price not verified yet.Price not verified yet.
Microsoft Azure8$23,925$4,421$45,935billed every hour
Google CloudPrice not verified yet.Price not verified yet.Price not verified yet.

Every offer and its price

Every listed offer for this GPU. Where a provider sells several instance sizes, the tables above use the cheapest size that fits the GPU count.

NVIDIA H100 80GB (SXM) offers
ProviderInstanceLocationGPUs per instanceOn demandSpot1-year reserved
AWS EC2p5.48xlargeUS East (N. Virginia), Linux8$55.04per instance-hourn/an/a
Microsoft AzureStandard_ND96isr_H100_v5East US, Linux8$98.32per instance-hour$18.169536per instance-hour$551,221.00per instance per 1-year term
Google Clouda3-highgpu-8gus-central18n/an/an/a
Lambda8x NVIDIA HGX H100Lambda Cloud8$4.59per GPU-hourNot offeredNot offered
Lambda4x NVIDIA H100 SXMLambda Cloud4$4.69per GPU-hourNot offeredNot offered
Lambda2x NVIDIA H100 SXMLambda Cloud2$4.79per GPU-hourNot offeredNot offered
Lambda1x NVIDIA H100 SXMLambda Cloud1$4.89per GPU-hourNot offeredNot offered
Runpod Secure CloudH100 SXMRunpod Secure CloudAny$3.99per GPU-hourNot offeredNot offered
Runpod Community CloudH100 SXMRunpod Community CloudAny$2.69per GPU-hourNot offeredNot offered

No verified throughput figure for this GPU yet, so cost per million tokens is not shown.

Your traffic looks different? Enter your own figures in the calculator.

Sources

Sources: aws.amazon.com (checked ); b0.p.awsstatic.com (checked ); lambda.com (checked ); learn.microsoft.com (checked ); prices.azure.com (checked ); runpod.io (checked )

Prices are list prices as each provider publishes them, used unchanged. How the figures are calculated: methodology.