A100-PCIE-40GB vs A100-SXM-80GB
A100-SXM-80GB is the cheaper of the two: $1.07/hr at Vast.ai, against $1.99/hr for A100-PCIE-40GB at Lambda — 1.9× the price.
Cheapest on each
Aligned facts
Same sheet, same order, both cards — including the gaps.
| Attribute | A100-PCIE-40GB | A100-SXM-80GB |
|---|---|---|
| VRAM | 40 GB | 80 GB▲ |
| Bandwidth | ○ no answer | 2.039 TB/s |
| FP16 | ○ no answer | 312 TFLOPS |
| FP8 | ○ no answer | ○ no answer |
| TDP | ○ no answer | 400 W |
| NVLink | ○ no answer | NVLink 3 |
| Architecture | Ampere | Ampere |
| Released | ○ no answer | 2020 |
A blank is not a zero. Where one card states a figure and the other states none, neither wins the row — the second number is unknown, and ranking an unknown as nothing is how a comparison starts to lie.
Every seller, both cards
One row per seller, cheapest tier, on-demand list prices only. Sellers of both cards lead, cheapest combined rate first.
| Seller | A100-PCIE-40GB | A100-SXM-80GB |
|---|---|---|
| Lambda | $1.992026-10-26 | $2.792026-10-26 |
| Vast.ai | ○ no answer | $1.072026-10-27 |
| Beam | ○ no answer | $1.362026-10-27 |
| Massed Compute | ○ no answer | $1.382026-10-26 |
| Hyperstack | ○ no answer | $1.602026-10-26 |
| RunPod | ○ no answer | $1.792026-10-27 |
| TensorDock | ○ no answer | $1.802026-10-26 |
| Koyeb | ○ no answer | $2.152026-10-26 |
| Verda (DataCrunch) | ○ no answer | $2.182026-10-27 |
What this page will not tell you
On-demand list rates only · spot and committed terms priced separately and not mixed in here · no performance-per-dollar figure appears on this page, because it needs a throughput benchmark we do not hold and dividing a price by a vendor’s own TFLOPS claim produces a number that looks authoritative and means nothing