The cheapest A100 SXM on the table right now is $0.130 per GPU-hour at Vast.ai (spot), out of 52 offers from 26 providers. The A100 SXM carries 80 GB of VRAM on NVIDIA's Ampere generation. That is peer/marketplace capacity — rented from other users' machines, and not directly comparable to a dedicated instance.
On-demand pricing for the same card spans $0.667 to $5.04 per GPU-hour — a 7.6× 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 $0.667 at Vast.ai · spot $0.130 at Vast.ai · reserved $0.995 at LeaderGPU.
| Provider | Tier | $/GPU-hr | Pricing | Node | Commit | In stock | |
|---|---|---|---|---|---|---|---|
| Vast.ai | marketplace | $0.130 | spot | 1–4× GPU | — | yes | rent |
| Google Cloud | dedicated | $0.410 | spot | 1× GPU | — | — | rent |
| Vast.ai | marketplace | $0.667 | on-demand | 4× GPU | — | yes | rent |
| Vast.ai | marketplace | $0.668 | on-demand | 2× GPU | — | yes | rent |
| Vast.ai | marketplace | $0.835 | on-demand | 1× GPU | — | yes | rent |
| Verda | dedicated | $0.895 | spot | 1–8× GPU | — | — | rent |
| LeaderGPU | dedicated | $0.995 | reserved | 8× GPU | 1 mo | — | rent |
| Azure | dedicated | $1.06 | spot | 8× GPU | — | — | rent |
| Thunder Compute | dedicated | $1.09 | on-demand | 1× GPU | — | yes | rent |
| GPU.ai | dedicated | $1.12 | on-demand | 1× GPU | — | yes | rent |
| Seeweb | dedicated | $1.15 | on-demand | 1× GPU | — | — | rent |
| Prime Intellect | dedicated | $1.23 | on-demand | 1× GPU | — | yes | rent |
| Hyperstack | dedicated | $1.35 | on-demand | 1× GPU | — | — | rent |
| Massed Compute | dedicated | $1.35 | on-demand | 1–8× GPU | — | — | rent |
| Runpod Community | marketplace | $1.39 | on-demand | 1× GPU | — | yes | rent |
| Runpod Community | marketplace | $1.39 | spot | 1× GPU | — | yes | rent |
| Spheron | dedicated | $1.43 | on-demand | 1× GPU | — | — | rent |
| Denvr Dataworks | dedicated | $1.45 | on-demand | 8× GPU | — | — | rent |
| Jarvis Labs | dedicated | $1.49 | on-demand | 1× GPU | — | yes | rent |
| Runpod | dedicated | $1.59 | on-demand | 1× GPU | — | yes | rent |
| Runpod | dedicated | $1.59 | spot | 1× GPU | — | yes | rent |
| Koyeb | dedicated | $1.60 | on-demand | 1× GPU | — | — | rent |
| Verda | dedicated | $1.79 | on-demand | 1–8× GPU | — | — | rent |
| Civo | dedicated | $1.79 | on-demand | 1–8× GPU | — | — | rent |
| AWS | dedicated | $2.15 | spot | 8× GPU | — | — | rent |
| Beam | dedicated | $2.25 | on-demand | 1× GPU | — | yes | rent |
| Crusoe | dedicated | $2.30 | on-demand | 8× GPU | — | — | rent |
| CoreWeave | dedicated | $2.70 | on-demand | 8× GPU | — | — | rent |
| Lambda Labs | dedicated | $2.79 | on-demand | 8× GPU | — | — | rent |
| AWS | dedicated | $3.43 | on-demand | 8× GPU | — | — | rent |
| OVHcloud | dedicated | $3.49 | on-demand | 1× GPU | — | — | rent |
| Google Cloud | dedicated | $3.93 | on-demand | 1× GPU | — | — | rent |
| Baseten | dedicated | $4.00 | on-demand | 1× GPU | — | yes | rent |
| Azure | dedicated | $4.10 | on-demand | 8× GPU | — | — | rent |
| Replicate | dedicated | $5.04 | on-demand | 1, 2× GPU | — | yes | rent |
| Replicate | dedicated | $5.04 | reserved | 4, 8× GPU | — | yes | rent |
How much does an A100 SXM cost per hour?
As of the latest scrape, $0.130 per GPU-hour at Vast.ai is the cheapest A100 SXM offer across 26 tracked cloud providers. The cheapest on-demand rate is $0.667 per GPU-hour at Vast.ai, and the most expensive on-demand rate is $5.04 at Replicate.
Which cloud is cheapest for the A100 SXM?
Vast.ai at $0.130 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 A100 SXM capacity cheaper?
Yes — the cheapest spot A100 SXM is $0.130 per GPU-hour at Vast.ai, -81% versus the cheapest on-demand rate of $0.667. 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 A100 SXM have?
80 GB, on the Ampere architecture. Multi-GPU instances multiply that: an 8× A100 SXM node exposes 640 GB of GPU memory in total.