The cheapest B300 on the table right now is $3.69 per GPU-hour at AWS (spot), out of 26 offers from 12 providers. The B300 carries 288 GB of VRAM on NVIDIA's Blackwell generation.
On-demand pricing for the same card spans $6.60 to $17.80 per GPU-hour — a 2.7× 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 $6.60 at Massed Compute · spot $3.69 at AWS · reserved $4.30 at Nebius.
| Provider | Tier | $/GPU-hr | Pricing | Node | Commit | In stock | |
|---|---|---|---|---|---|---|---|
| AWS | dedicated | $3.69 | spot | 8× GPU | — | — | rent |
| Verda | dedicated | $3.75 | spot | 1–8× GPU | — | — | rent |
| Nebius | dedicated | $4.30 | reserved | 8× GPU | — | — | rent |
| fal | dedicated | $4.49 | reserved | 1× GPU | — | — | rent |
| Spheron | dedicated | $5.81 | spot | 1× GPU | — | — | rent |
| Massed Compute | dedicated | $6.60 | on-demand | 8× GPU | — | — | rent |
| Modal | dedicated | $7.10 | on-demand | 1× GPU | — | yes | rent |
| Hyperstack | dedicated | $7.40 | on-demand | 1× GPU | — | — | rent |
| Verda | dedicated | $7.50 | on-demand | 1–8× GPU | — | — | rent |
| Nebius | dedicated | $7.85 | on-demand | 8× GPU | — | — | rent |
| Runpod | dedicated | $7.89 | on-demand | 1× GPU | — | yes | rent |
| Runpod | dedicated | $7.89 | spot | 1× GPU | — | yes | rent |
| DigitalOcean | dedicated | $7.94 | reserved | 8× GPU | 12 mo | — | rent |
| fal | dedicated | $8.50 | on-demand | 1× GPU | — | — | rent |
| Lium.io | marketplace | $8.50 | on-demand | 1× GPU | — | yes | rent |
| Scaleway | dedicated | $8.73 | on-demand | 8× GPU | — | — | rent |
| Scaleway | dedicated | $9.92 | on-demand | 4× GPU | — | — | rent |
| Scaleway | dedicated | $11.04 | on-demand | 2× GPU | — | — | rent |
| DigitalOcean | dedicated | $11.19 | spot | 8× GPU | — | — | rent |
| AWS | dedicated | $17.80 | on-demand | 8× GPU | — | — | rent |
How much does a B300 cost per hour?
As of the latest scrape, $3.69 per GPU-hour at AWS is the cheapest B300 offer across 12 tracked cloud providers. The cheapest on-demand rate is $6.60 per GPU-hour at Massed Compute, and the most expensive on-demand rate is $17.80 at AWS.
Which cloud is cheapest for the B300?
AWS at $3.69 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 B300 capacity cheaper?
Yes — the cheapest spot B300 is $3.69 per GPU-hour at AWS, -44% versus the cheapest on-demand rate of $6.60. 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 B300 have?
288 GB, on the Blackwell architecture. Multi-GPU instances multiply that: an 8× B300 node exposes 2304 GB of GPU memory in total.