Rent NVIDIA A100 SXM (40GB) in the Cloud

The best-priced cloud NVIDIA A100 SXM (40GB) rate we track is $0.80 per hour on Vast.ai. That rate reflects on-demand billing; providers that support spot/preemptible instances bring the floor down to per hour — a saving of around up to 60% for workloads that can tolerate interruption.

If you can commit to reserved capacity, expect a further up to 40% off. For ad-hoc experiments we recommend on-demand; for multi-day pre-training budget a mix of spot and on-demand. Hourly quotes can shift weekly as providers compete, so check the live table before committing a long run.

Launch a NVIDIA A100 SXM (40GB) instance on Vast.ai at $0.80/hr, or try RunPod for alternative regions and availability.

At a glance: NVIDIA A100 SXM (40GB) = Ampere architecture, 40 GB HBM2e memory, 1,555 GB/s bandwidth, 312 FP16 TFLOPS, 19.5 FP32 TFLOPS, 400W, 2020.

Those specs tell most of the story for machine learning: VRAM sets the model ceiling, bandwidth throttles attention-heavy production inference, and TFLOPS set pre-training throughput. NVIDIA A100 SXM (40GB) sits firmly in the class of accelerators targeted at modern transformer workloads — the bandwidth/TFLOPS balance is tuned for large-batch pre-training and production production inference rather than gaming.

Two tracked cloud providers currently offer NVIDIA A100 SXM (40GB): Vast.ai and RunPod. Vast.ai has the cheaper rate at $0.80/hr.

Real performance numbers for NVIDIA A100 SXM (40GB): 312 FP16 TFLOPS / 19.5 FP32 TFLOPS / 1,555 GB/s bandwidth / 40 GB VRAM.

For context, that's the compute envelope used to train and serve most commercial LLMs in production today. Whether you're running research-grade training or latency-sensitive low-latency inference, NVIDIA A100 SXM (40GB) generally hits the throughput envelope expected from its Ampere generation, with the specific number driven by batch size, sequence length, and quantisation choices.

Deploy NVIDIA A100 SXM (40GB) on Vast.ai (from $0.80/hr) or RunPod — check live availability and spin up in minutes.

The 2 providers offering NVIDIA A100 SXM (40GB) today: RunPod, Vast.ai. The headline-cheapest on-demand source is Vast.ai at $0.80 per hour; spot pricing bottoms out at per hour on .

Real-world selection factors include: interconnect technology (NVLink vs. PCIe vs. InfiniBand), multi-node availability, per-second vs. per-hour billing, egress fees, and regional coverage. A cheap provider with high egress fees can cost more than a pricier provider with free egress once you factor in data movement.

Rent NVIDIA A100 SXM (40GB) from Vast.ai (from $0.80/hr) or RunPod — compare live pricing and deploy.

What is NVIDIA A100 SXM (40GB) good for? AI training, fine-tuning, inference on smaller models. Those are the natural fits given its specs: enough memory (40 GB) for most modern LLMs, enough compute for training at scale, and enough bandwidth to keep inference latency low.

NVIDIA A100 SXM (40GB) is not the cheapest option for small models, nor the largest option for frontier-scale training — but it occupies the highest-volume middle of the AI accelerator market, which is why it appears in most cloud catalogues at $0.80 per hour or better.

Rent NVIDIA A100 SXM (40GB) from Vast.ai (from $0.80/hr) or RunPod — compare live pricing and deploy.