H100 PCIe Cloud GPU Rental - Compare 10 Providers

Find the cheapest H100 PCIe cloud GPU rental. Prices range from $1.99/hr to $4.00/hr across 10 providers.

Quick answer: The cheapest H100 PCIe rentals are usually on lower‑cost providers, but reliability and region matter. Use the table below to compare price, availability, and region.
10
Providers
$1.99/hr
Cheapest
$2.74/hr
Average
80 GB
VRAM

Price Trend

Window Min Avg Max Start → End
Last 7 days$1.99$1.99$1.99$1.99 → $1.99
Last 30 days$1.99$1.99$1.99$1.99 → $1.99
Last 90 days$1.99$1.99$1.99$1.99 → $1.99

Provider Comparison

Rank Provider GPU VRAM Price/Hr Region Availability Link
1 RunPod H100 PCIe 80 GB $1.99/hr Global High Rent →
2 FluidStack H100 PCIe 80 GB $2.35/hr Global Medium Rent →
3 Shadeform H100 PCIe 80 GB $2.35/hr Global High Rent →
4 Lambda H100 PCIe 80 GB $2.49/hr US/EU Very Low Rent →
5 MassedCompute H100 PCIe 80 GB $2.49/hr US Medium Rent →
6 Hyperstack H100 PCIe 80 GB $2.65/hr US/EU High Rent →
7 Nebius H100 PCIe 80 GB $2.80/hr EU/Israel High Rent →
8 DataCrunch H100 PCIe 80 GB $2.99/hr EU High Rent →
9 Scaleway H100 PCIe 80 GB $3.27/hr EU Medium Rent →
10 CoreWeave H100 PCIe 80 GB $4.00/hr US/EU Medium Rent →

Cost Calculator

Duration RunPodFluidStackShadeform
1 hour$1.99$2.35$2.35
6 hours$11.94$14.10$14.10
1 day (24h)$47.76$56.40$56.40
1 week (168h)$334.32$394.80$394.80
1 month (730h)$1452.70$1715.50$1715.50

💡 Pro Tips for Renting H100 PCIe

Related Guides

Compare Other Cloud GPUs

H100 A100 80GB A100 40GB RTX 4090 RTX 3090 RTX A6000 L40S

Looking to buy instead of rent? Check out our GPU buying guide or read the FAQ for more tips.

FAQ

How much does it cost to rent H100 PCIe in the cloud?
H100 PCIe rentals range from $1.99/hr to $4.00/hr depending on provider, region, and availability.
Why do prices vary so much between providers?
Pricing differences come from reliability tier, region, contract type (spot vs on‑demand), and hardware availability.
Which provider should I choose for production?
Choose providers with stronger reliability and uptime SLAs if you are running long training or production workloads.
Is the cheapest option always best?
Cheapest is best for experiments. For production or long runs, prioritize reliability and availability.