Home / Research / GPU Cloud Pricing Statistics

Market research · August 29, 2026

GPU Cloud Pricing Statistics: September 2026

By Darin ten Bruggencate, Founder, AI Industry Reviews

Renting a data center GPU can cost very different amounts depending on the provider and how the capacity is sold. Teams deciding whether to rent or buy need comparable prices.

Our analysts collected 4,180 GPU-hour rental prices for 14 data center accelerators across 61 cloud providers. We gathered them between March 2 and August 29, 2026. The sample covers the three largest hyperscalers, 22 specialist GPU clouds and the major peer-to-peer marketplaces. We converted every listing to the price of a single GPU-hour so all offers sit on the same scale.

In the following sections, we compare H100 prices by provider type, show when owning beats renting, and look at price spreads, trends and commitment discounts.

H100 Pricing by Provider Category

In the table below, we compare median on-demand rates across three provider categories and express the H100 gap as a premium over the marketplace baseline.

H100 Pricing by Provider Category, September 2026

Insights

• Hyperscaler H100 on-demand priced at a 170% premium to marketplace in the September sample, or $5.64 versus $2.09 per GPU-hour.
• Specialist neocloud sat between the two at $3.28, a 57% premium to marketplace.
• For AI teams renting GPUs, where they rent matters as much as what they rent. A hyperscaler H100 cost $5.64 an hour, against $3.28 at neoclouds and $2.09 on marketplaces. Price-sensitive buyers are likely to keep shifting work away from hyperscalers.

The hyperscaler rate buys a support contract, a service level agreement, private networking into an existing cloud footprint, and compliance attestations that most specialist providers do not offer. We sorted providers by those attributes rather than by company size, which is why two of the larger specialist fleets in the sample are in the neocloud row.

Owning Versus Renting: Break-Even Utilization

In the table below, we model the all-in cost of owning each accelerator per GPU-hour over a five-year life at 85% utilization, then solve for the utilization rate at which ownership matches renting the same card on demand at the sample median.

Owning Versus Renting Break-Even, September 2026

Insights

• An H100 purchase only beats median on-demand rental once the card stays busy at least 50% of the time.
• B200 breaks even at 38% utilization, the lowest threshold in the set, because rental medians remain elevated relative to street price.
• Buying GPUs only pays off for teams that keep them busy. An H100 beats renting at 50% utilization, and a B200 at 38%. Companies without steady workloads are likely to do better renting.

Price Spread by GPU Model

List price for a single provider is a narrow slice of the market. In the table below, we report the sample median on-demand rate for each model plus the 10th and 90th percentile of listings. The spread column is the 90th percentile divided by the 10th, which measures how fragmented pricing is for that card.

GPU Price Spread by Model, September 2026

Insights

• H100 showed the widest spread in the sample at 6.2x, reflecting its dual life as premium enterprise capacity and commodity overflow.
• V100 rented above L40S in half of the markets sampled, which tracks legacy CUDA pin rather than performance.
• Prices for the same chip vary widely, so shopping around pays. H100 rates spread 6.2 times between the 10th and 90th percentiles, and MI300X sat 33% below H200. Buyers open to AMD hardware are likely to find the deepest discounts.

H100 Prices by Provider Channel, Q3 2023 to Q3 2026

In the table below, we show the median quarterly on-demand H100 rate in each of the three provider channels.

H100 Prices by Provider Channel, Q3 2023 to Q3 2026

In the quarterly series above, hyperscaler pricing rose through 2024 while reserved enterprise capacity stayed scarce, then fell in mid-2025 as specialist competition forced Hopper fleets to reprice. Marketplace rates fell about 50% from the third quarter of 2023 to the fourth quarter of 2025 as resellers and colocation tenants listed idle cards. AI Industry Reviews reads the 2026 uptick across all three channels as the first sustained reversal in that series, tied to high-bandwidth memory costs passing into rental rates.

Commitment Discounts Versus On-Demand

In the table below, we quantify the median discount each commitment structure earns against the same provider's on-demand rate, plus the effective hourly rate a three-year commitment produces.

GPU Commitment Discounts, September 2026

Insights

• Spot discounts widened with hardware age, from 31% on B200 to 63% on L40S.
• One-year reserved discounts stayed smaller than spot discounts on every model priced, so they are the weakest value for interruptible work.
• Long commitments now beat on-demand marketplace prices. A three-year H100 commitment prices below the median marketplace on-demand rate, a first in this series. Teams with predictable demand are likely to save most by committing.

What GPU Cloud Pricing Trends Mean for 2027 and Beyond

Renting AI chips has become much cheaper for older models, while the newest chips still cost a lot because supply is short. Where a team rents and how long it commits for now shape the bill more than the chip itself.

As newer chips arrive in volume, prices for current ones should keep falling, and the premium on the newest ones should shrink. Buying hardware outright only pays off for teams that keep it busy most of the time, so most teams are better off renting until their workload is steady.

So if you rent GPUs, match the contract to your workload: commit long only for steady jobs, and use marketplaces or spot capacity for everything else.

Requesting a Copy of This Report

If you would like a PDF copy of this report, or the underlying dataset of price observations, you can reach out here.