Search the market

Jump to a section, a GPU or a provider

Live market· updated Fri 11:00

The live market for GPU compute.

Every GPU cloud on one honest board - hyperscalers, neoclouds and EU-sovereign providers, side by side in euros. See what compute really costs this hour, when the grid makes it cheaper, and where your data stays under European law.

27 providers · 420 configurations · prices refreshed with every grid hour

Cheapest NVIDIA H200 141GB ×1 right now: €2.01/hr on Leafcloud. Market median €4.12/hr.

Dutch grid, this hour€-/MWh
€0 above 30-day avg
Low
€160 · Fri 01:00
High
€250 · Fri 16:00
Dynamic GPU rates
0% vs base

Prices from 27 GPU clouds

  • Leafcloud
  • OVHcloud
  • Scaleway
  • Nebius AI Cloud
  • Verda (formerly DataCrunch)
  • STACKIT
  • IONOS Cloud
  • Hetzner Online
  • Exoscale
  • evroc
  • Hyperstack (NexGen Cloud)
  • Amazon Web Services
  • Microsoft Azure
  • Google Cloud
  • OCI
  • IBM Cloud
  • CoreWeave
  • Lambda Labs
  • Crusoe Cloud
  • Together AI
  • Voltage Park
  • DigitalOcean GPU Droplets
  • Paperspace
  • Vultr
  • Hot Aisle
  • Massed Compute
  • RunPod

Price spread

2.2×

€2.14 to €4.91 per hour across 6 listings

Data sovereignty

31% EU-sovereign

2 of 6 listings sit outside US CLOUD Act reach

  • EU-sovereign2
  • EU · US subsidiary1
  • US jurisdiction2
  • Marketplace1

Dynamic pricing this hour

-

Waiting for the grid feed

−50% · cheap grid+50% · tight grid

Next 0 hours on the grid

-

Tomorrow's prices publish around 13:00

Price over the next 24 hours

6 listings for this configuration. Every hour is a real day-ahead price.

GPU
GPUs per instance
Run length
Show prices as
Dynamic - follows the gridStatic market range, dashed median

Every NVIDIA H200 141GB ×1 listing, ranked

Live prices in euros. Filter by where your data sits and how the provider prices; click a row for the full quote.

Sort by
#ProviderConfigurationData residencyPricingPer hour
1
LeafcloudCheapest right now
H200 ×1Specs not published
EU-sovereignams-1
Dynamic +3%
€2.14
-48% vs median
Book
2
RunPod
H200 · Community24 vCPU · 276 GB RAM
Marketplaceglobal
Flat rate
€3.27
-21% vs median
Book
3
DigitalOcean GPU Droplets
gpu-h200x1-141gb24 vCPU · 240 GB RAM · 5.1 TB NVMe
USNYC2
Flat rate
€4.07
-1% vs median
Book
4
RunPod
H200 · Secure24 vCPU · 276 GB RAM
USglobal
Flat rate
€4.18
+1% vs median
Book
5
Verda (formerly DataCrunch)
1x H200 SXM5 141GB44 vCPU · 170 GB RAM
EU-sovereignFI
Flat rate
€4.35
+6% vs median
Book
6
Nebius AI Cloud
gpu-h200-sxm 1gpu-16vcpu-200gb16 vCPU · 200 GB RAM
EU · US subeu-north1
Flat rate
€4.91
+19% vs median
Book

Browse by GPU

35 chips on the market. Bars show each chip's €/GPU-hr range across providers, with the median marked.

Top-end training silicon. H100/H200, MI300X, Blackwell. Highest cost, highest throughput.

The provider landscape

All 27 providers by what they charge chip-for-chip against the market median, and how well your data is protected. Click a provider to see everything it sells.

Head-to-head

Six scores from 0 to 100, each relative to the rest of the market - a bigger shape is a stronger all-round offer. Press on any provider in the leaderboard below to add it (up to 3); hover the chart to read each score and what it measures.

  • Voltage Park
  • Leafcloud
  • AWS

Every provider, ranked

Ordered by price index: 1.00× is the market median for the same chips, lower is cheaper. Press to put a provider in the head-to-head above, or click its name for compliance and every configuration it sells.

Sort by
  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  1. 15
  2. 16
  3. 17
  4. 18
  5. 19
  6. 20
  7. 21
  8. 22
  9. 23
  10. 24
  11. 25
  12. 26
  13. 27
Bar = price vs market median, chip for chip (center = market) adds to the radar (up to 3)Dot = where your data sits (sovereignty)