A100 40GB
See rental prices ↓Current median: $1.42/hr across 22 providers (observed), 0.00 (0.0%) vs yesterday. Typical range: $0.96–$6.05/hr (P10–P90). 30-day trend unlocks after 30 days of captures
TDP 300W · 312 TFLOPS · 40GB VRAM · 1555 GB/s bandwidth
The A100 40GB has a 300W TDP, 312 TFLOPS of FP16 compute, 40GB of HBM2e VRAM, 1555 GB/s memory bandwidth.
Same compute as 80GB variant but half the VRAM. Lower cost, suitable for models up to ~30B params.
A100 40GB currently rents for a median of $1.42/hr across 22 providers (observed on-demand prices). A100 40GB has 40GB of memory. Compare its specs and listed cloud rental prices below. The workload must fit in VRAM with room for runtime memory; a listed price does not confirm stock.
A100 40GB Overview
The A100 40GB is the smaller-memory version of NVIDIA’s Ampere workhorse, retaining the GA100 Tensor Core design and 312 TFLOPS of FP16/BF16 throughput. It offers 40GB of HBM2e, 1,555 GB/s of bandwidth, and NVLink 3.0 support, making it a mature option for research and production workloads that do not need 80GB per card.
Forty gigabytes can accommodate roughly a 20B-parameter model at FP16 before allowing for activations and cache, while 1,555 GB/s gives it far more inference bandwidth than GDDR-based cards in its price band. NVLink 3.0 can connect cards for distributed training or model parallelism, though a single card cannot make a larger model fit by itself.
The A100 40GB is well suited to mid-size training, established CUDA pipelines, and inference where Ampere software maturity matters. It lacks Hopper’s FP8 Transformer Engine, and its lower bandwidth and capacity make it less compelling for large-context or very large-model serving.
Memory
Compute Performance
Hardware Specifications
Chip
Processors
Clocks
Memory
Power
Relative Performance
Relative to highest-spec GPU in database
Limitations
Live Cloud PricingOn-demand hourly rates
| Provider | On-Demand | Spot | Region | Availability |
|---|---|---|---|---|
BESTPaperspace | $7.90/hr | — | us-east-1 | unknown |
| $0.90/hr | — | US | unknown | |
| $1.98/hr | — | IN | unknown | |
| $1.42/hr | — | US/EU | unknown | |
| $2.19/hr | — | US | unknown | |
| $3.58/hr | — | cn-east-1 | unknown | |
| $5.60/hr | — | EU/US | unknown | |
| $1.10/hr | — | EU-West | unknown | |
| $1.45/hr | — | US | unknown | |
| $1.35/hr | — | US-West | unknown | |
| $1.20/hr | — | US | unknown | |
| $1.10/hr | — | US | unknown | |
| $3.67/hr | — | us-central1 | unknown | |
| $1.29/hr | — | US | unknown | |
| $1.10/hr | — | US | unknown | |
| $6.10/hr | — | US/EU/BR | unknown | |
| $1.93/hr | — | Phoenix, AZ, United States, Dulles, VA, United States, New York, NY, United States, Calgary, Canada, Helsinki, Finland, San Jose, CA, United States | unknown | |
| $1.09/hr | — | us-east-1 | unknown | |
| $2.10/hr | — | us-east | unknown | |
| $1.99/hr | — | us-west-1 | unknown | |
| $7.30/hr | — | US/EU/APAC | unknown | |
| $4.90/hr | — | US/EU/APAC | unknown | |
| $0.95/hr | — | US | unknown | |
| $0.65/hr | — | eu-north1 | unknown | |
| $0.65/hr | — | EU/US | unknown | |
| $1.60/hr | — | EU-Central | unknown | |
| $10.24/hr | — | us-east-1 | unknown | |
| $0.29/hr | — | EU-South | unknown | |
| $1.56/hr | — | EU-North | unknown |
Job Cost Estimator
At median $1.42/hr across 22 providers (observed on-demand). Spot pricing typically 40–70% lower.
Estimates only. Actual cost depends on provider, region, availability, and utilization. Use the GPU cost calculator for custom inputs.
Compare A100 40GB vs…
Use Case Guidance
LLM Model Size Guidance
Estimates only. Actual capacity depends on context length, KV cache, and framework overhead.
Related Guides
LLM APIs Running on This GPU Class
Providers that serve frontier LLM inference on Ampere-class hardware.
A100 40GB vs Alternatives — Spec Comparison
| Spec | A100 40GB this | A100 80GB | A100 SXM4 80GB | A100 PCIe 80GB |
|---|---|---|---|---|
| VRAM | 40GB HBM2e | 80GB HBM2e▲ | 80GB HBM2e▲ | 80GB HBM2e▲ |
| Memory Bandwidth | 1555 GB/s | 2000 GB/s▲ | 2000 GB/s▲ | 1935 GB/s |
| FP16 TFLOPS | 312▲ | 312▲ | 312▲ | 312▲ |
| BF16 TFLOPS | 312▲ | 312▲ | 312▲ | 312▲ |
| FP8 TFLOPS | — | — | — | — |
| INT8 TOPS | 624▲ | 624▲ | 624▲ | 624▲ |
| TDP | 300W | 400W▲ | 400W▲ | 300W |
| Process Node | TSMC 7nm▲ | TSMC 7nm▲ | TSMC 7nm▲ | TSMC 7nm▲ |
| Architecture | GA100▲ | GA100▲ | GA100▲ | GA100▲ |
| Release Year | 2020 | 2020 | 2020 | 2021▲ |
| Max model (FP16) | ~20B params | ~40B params▲ | ~40B params▲ | ~40B params▲ |
| Max model (INT4) | ~80B params | ~160B params▲ | ~160B params▲ | ~160B params▲ |
Related GPUs
A100 40GB Frequently Asked Questions
What is the A100 40GB TDP?
The A100 40GB has a TDP (Thermal Design Power) of 300W.
How many TFLOPS does the A100 40GB have?
The A100 40GB delivers 312 TFLOPS of FP16 compute and 312 TFLOPS BF16.
How much VRAM does the A100 40GB have?
The A100 40GB has 40GB of HBM2e VRAM.
What is the A100 40GB's memory bandwidth?
The A100 40GB has a memory bandwidth of 1555 GB/s.
A100 40GB User Reviews
A100 40GB User Reviews
Ready to rent?
Compare A100 40GB prices across 130+ providers
Live on-demand & spot rates · monthly cost estimates · availability status