NVIDIA A100 40GB rental price
DatacenterThe original A100. Deep supply at the hyperscalers and steadily falling rates as Hopper capacity comes online.
live · updated 17d ago
VRAM
40 GB HBM2
Architecture
Ampere
Segment
Datacenter
Providers quoting
9
best rate
$0.536/GPU-hr at Vast.ai
24h
—
7d
—
30d
—
cheapest 8× node
$11.20/hr
The cheapest NVIDIA A100 40GB you can rent right now is $0.536 per GPU-hour at Vast.ai, against $3.400 at the dearest of the 9 providers quoting it — a 6.3× spread for identical silicon. A full 8-GPU node starts at $11.20/hr at Denvr Dataworks. Rates below refresh automatically.
Rent vs. buy
buy outright
$4,475
rent
$0.536/hr
breakeven
11.4 months
Buying a NVIDIA A100 40GB PCIe at today’s street price of $4,475 costs the same as 8,356 hours (11.4 months) of renting one at $0.536 an hour. Run it less than that and renting is simply cheaper; run it more and the card starts paying itself back — but this is a floor, not a total cost of ownership. It counts only the purchase price against the rental rate, and ignores power, cooling, rack space, networking, depreciation and the cost of tying up the capital, all of which push the real breakeven further out.
NVIDIA A100 40GB PCIe purchase prices and history →A100-40 hourly rate history
Daily low $/GPU-hour per provider, cheapest provider first — hover the chart for detail.
Where to rent the NVIDIA A100 40GB
Vast.ai
Marketplace
$0.536/GPU-hr
1× A100-40 · $0.536/hr per node · California, US
- 1× A100 SXM4 x1$0.601/hr
- 1× A100 SXM4 x1$0.601/hr
- 1× A100 SXM4 x1$0.602/hr
Verda
GPU cloud
$1.360/GPU-hr
1× A100-40 · $1.360/hr per node · sxm4 · FI, Central
no capacity right now
- 8× 8A100.40S.176V · no capacity$10.88/hr
Denvr Dataworks
GPU cloud
$1.400/GPU-hr
8× A100-40 · $11.20/hr per node · sxm4 · CA, Calgary
- 1× A100_40GB_SXM_1x$1.400/hr
Crusoe
GPU cloud
$1.450/GPU-hr
1× A100-40 · $1.450/hr per node · pcie · US, Fairview, OH
no capacity right now
- 8× a100.8x · no capacity$11.60/hr
- 4× a100.4x · no capacity$5.800/hr
- 2× a100.2x · no capacity$2.900/hr
Lambda
GPU cloud
$1.990/GPU-hr
1× A100-40 · $1.990/hr per node · sxm4 · IN, Mumbai
- 1× gpu_1x_a100 · no capacity$1.990/hr
- 4× gpu_4x_a100 · no capacity$7.960/hr
- 8× gpu_8x_a100$15.92/hr
Paperspace
GPU cloud
$3.190/GPU-hr
1× A100-40 · $3.190/hr per node · pcie · US, New York, NY
- 4× A100x4 · no capacity$12.44/hr
- 2× A100x2 · no capacity$6.260/hr
- 1× A100$3.190/hr
Microsoft Azure
Hyperscaler
$3.400/GPU-hr
8× A100-40 · $27.20/hr per node · eastus
- 8× ND96asr_v4$27.20/hr
A100-40 cluster nodes
Multi-GPU configurations, cheapest per GPU-hour first. For distributed training the interconnect matters as much as the rate — an NVLink or InfiniBand node and the same count of loose PCIe cards are not the same machine.
| Provider | Configuration | Interconnect | $/GPU-hr | Node/hr |
|---|---|---|---|---|
| Verdano capacity | 8× 8A100.40S.176V · FI, Central | sxm4 | $1.360 | $10.88/hr |
| Denvr Dataworks | 8× A100_40GB_SXM_8x · CA, Calgary | sxm4 | $1.400 | $11.20/hr |
| Crusoeno capacity | 8× a100.8x · US, Fairview, OH | pcie | $1.450 | $11.60/hr |
| Crusoeno capacity | 4× a100.4x · US, Fairview, OH | pcie | $1.450 | $5.800/hr |
| Crusoeno capacity | 2× a100.2x · US, Fairview, OH | pcie | $1.450 | $2.900/hr |
| Lambdano capacity | 4× gpu_4x_a100 · IN, Mumbai | pcie | $1.990 | $7.960/hr |
| Lambda | 8× gpu_8x_a100 · US, San Jose, Ca | pcie | $1.990 | $15.92/hr |
| Lambdano capacity | 2× gpu_2x_a100 · IN, Mumbai | pcie | $1.990 | $3.980/hr |
| AWS | 8× p4d.24xlarge · us-east-1 | — | $2.745 | $21.96/hr |
| Paperspaceno capacity | 8× A100x8 · NL, Amsterdam | pcie | $3.110 | $24.88/hr |
| Paperspaceno capacity | 4× A100x4 · NL, Amsterdam | pcie | $3.110 | $12.44/hr |
| Paperspaceno capacity | 2× A100x2 · NL, Amsterdam | pcie | $3.130 | $6.260/hr |