The cheapest GB200 on the table right now is $6.22 per GPU-hour at Azure (spot), out of 4 offers from 3 providers. The GB200 carries 186 GB of VRAM on NVIDIA's Blackwell generation.
On-demand pricing for the same card spans $10.50 to $27.04 per GPU-hour — a 2.6× spread between the cheapest and the most expensive provider for identical silicon. That gap is the entire reason this table exists.
Cheapest by pricing model: on-demand $10.50 at CoreWeave · spot $6.22 at Azure.
| Provider | Tier | $/GPU-hr | Pricing | Node | Commit | In stock | |
|---|---|---|---|---|---|---|---|
| Azure | dedicated | $6.22 | spot | 4× GPU | — | — | rent |
| CoreWeave | dedicated | $10.50 | on-demand | 4× GPU | — | — | rent |
| Oracle Cloud | dedicated | $16.00 | on-demand | 4× GPU | — | — | rent |
| Azure | dedicated | $27.04 | on-demand | 4× GPU | — | — | rent |
How much does a GB200 cost per hour?
As of the latest scrape, $6.22 per GPU-hour at Azure is the cheapest GB200 offer across 3 tracked cloud providers. The cheapest on-demand rate is $10.50 per GPU-hour at CoreWeave, and the most expensive on-demand rate is $27.04 at Azure.
Which cloud is cheapest for the GB200?
Azure at $6.22 per GPU-hour on spot capacity. Prices move constantly, so this page is regenerated from the live scrape every 15 minutes. Note that marketplace tiers are peer hardware and are not directly comparable to dedicated capacity.
Is spot GB200 capacity cheaper?
Yes — the cheapest spot GB200 is $6.22 per GPU-hour at Azure, -41% versus the cheapest on-demand rate of $10.50. Spot instances can be reclaimed by the provider at any time, so they suit checkpointed training and batch inference rather than long-lived services.
How many GB of VRAM does the GB200 have?
186 GB, on the Blackwell architecture. Multi-GPU instances multiply that: an 8× GB200 node exposes 1488 GB of GPU memory in total.