GPU prices explained
How cloud GPU prices are quoted: price per GPU-hour, pricing models, configuration, location and availability.
Published Oct 9, 2026
Price per GPU-hour
Cloud providers quote GPU capacity per GPU and hour. A server with eight GPUs costs eight times the GPU-hour price per hour.
Storage, outbound data transfer and support are usually charged separately.
Pricing models
On-demand: billed by the hour, no commitment. Reserved: fixed capacity for one or three years at a lower price per hour.
Spot: unused capacity at a discount that the provider can reclaim. Community and marketplace: capacity from third parties, with varying reliability.
Location, currency and availability
The same GPU costs different amounts in different regions. Providers quote in different currencies, for example USD, EUR or CHF.
A list price does not guarantee available capacity. AI Compute Prices shows availability where the source reports it.
Spread in the market
For the H100, on-demand prices range from USD 2.64EUR 2.36CHF 2.18 to USD 13.27EUR 11.86CHF 10.94 per GPU-hour, with a median of USD 3.75EUR 3.35CHF 3.09, as of Oct 25, 2026.
As of Oct 25, 2026, 21:05 UTCSource: public price lists and pricing APIs of the providersConverted at ECB reference rates of Oct 23, 2026Method