Two Nvidia data center GPUs, one Ada Lovelace and one Pascal: 48 GB GDDR6 against 16 GB HBM2, 18% more memory bandwidth on the L40, and NVLink 160 GB/s on the P100 where the L40 is PCIe only.
At a glance
Memory
- L40
- 48 GB
- P100
- 16 GB
3x more on L40
Memory bandwidth
- L40
- 864 GB/s
- P100
- 732 GB/s
1.2x higher on L40
Cheapest on-demand
- L40
- $0.78 /GPU/hr
- P100
- $1.27 /GPU/hr
39% cheaper on L40
Price comparison
Median price per GPU per hour by billing type, with the cheapest listing under it.
| Billing | L40 | P100 | Difference |
|---|---|---|---|
| On-demand | $1.05 /GPU/hr from $0.78 | $1.53 /GPU/hr from $1.27 | 32% cheaper on L40 |
| Reserved | -- /GPU/hr from $0.70 | $1.04 /GPU/hr from $0.12 (24 mo) | |
| Spot | $0.79 /GPU/hr from $0.77 | -- /GPU/hr from $0.77 |
Heads up: A provider's own page may quote a different figure, for example on tax, region or a promotion. A monthly-only plan shows a derived hourly rate. Medians are across providers, one vote each, so a large catalog does not outweigh a small one. No median for reserved on the L40 and spot on the P100: we only show one when at least 3 providers list the GPU on that billing type. Verify before provisioning. More on how we price.
Price history
L40 vs P100 price history
Weekly median on-demand price per GPU per hour
Specs comparison
| Spec | L40 | P100 | Difference |
|---|---|---|---|
| Memory | |||
| Capacity | 48 GB GDDR6 | 16 GB HBM2 | 3x more on L40 |
| Bandwidth | 864 GB/s | 732 GB/s | 1.2x higher on L40 |
| Bus width | 384-bit | 4,096-bit | 11x wider on P100 |
| Compute · dense | |||
| FP8 | 362 TFLOPS | Not supported | L40 only |
| FP16 / BF16 | 181.1 TFLOPS | 18.7 TFLOPS | 9.7x higher on L40 |
| INT8 | 362 TOPS | Not supported | L40 only |
| FP32 | 90.5 TFLOPS | 9.3 TFLOPS | 9.7x higher on L40 |
| FP64 | Not supported | 4.7 TFLOPS | P100 only |
| Platform | |||
| Interconnect | PCIe only | NVLink 160 GB/s | NVLink on P100 only |
| Board power | 300 W | 250 W | 1.2x higher on L40 |
| Released | Q4 2022 | Q2 2016 | 6 years newer on L40 |
| Precision | |||
| Supported formats | FP8FP16BF16TF32FP32INT4INT8 | FP16FP32FP64 | |
Figures are for each model's headline configuration. Where a form factor carries a different number, it is named under the figure.
Compute: Vendor peak figures, dense and per GPU. Not measured throughput.
Source: L40 datasheet and P100 datasheet.
Where to rent them
Providers listing these GPUs, those with both first, so you can switch without moving clouds.
| Provider | L40 /GPU/hr | P100 /GPU/hr |
|---|---|---|
GPU.ai
|
On-Demand $0.78 | -- |
Thunder Compute
|
On-Demand $0.79 | -- |
Massed Compute
|
On-Demand $0.86 | -- |
Sesterce
|
On-Demand $0.97 | -- |
Runcrate
|
On-Demand $0.97 | -- |
Oblivus
|
On-Demand $1.05 | -- |
Spheron
|
On-Demand $1.06 | -- |
EmpirioLabs AI
|
On-Demand $1.50 | -- |
Scaleway
|
-- | On-Demand $1.42 |
Oracle Cloud
|
-- | On-Demand $1.27 |
Google Cloud
|
-- | On-Demand $1.51 |
Alibaba Cloud
|
-- | On-Demand $1.55 |
Database Mart
|
-- | Reserved (24mo) $0.12 |
12 providers list the L40 and 5 the P100. Every listing is on the L40 page and the P100 page.
Similar GPUs
Other comparisons in the same class.
Our data for Nvidia L40 was last updated on Sept. 7, 2026, and for Nvidia P100 on Sept. 7, 2026.