Hyperscalers US

Amazon Web Services.

EC2 GPU instances (p4, p5, g5, g6) on the largest cloud network. Reserved + Savings Plan discounts available; spot prices fluctuate by region. Best fit when you need broader AWS services (S3, SageMaker, VPC peering) alongside the GPU.

At a glance

Business model: Hyperscalers
Tier: Hyperscaler
Headquarters: US

When to pick Amazon Web Services

Best for

Enterprise customers already on AWS/GCP/Azure/Oracle.
Multi-region deployments and complex compliance setups.
Workloads tightly coupled to other cloud services (S3, BigQuery, etc.).

Avoid for

Cost-sensitive workloads — hyperscaler pricing is the highest tier.
Anyone who doesn't already need the surrounding cloud platform.

Price history

Daily median on Amazon Web Services's top GPUs.

GPUs available on Amazon Web Services

GPU	Tier	VRAM	$/hr	Range	Offers
Nvidia T4	datacenter	16GB	$0.53/hr	$0.53/hr–$0.53/hr	1	Compare →
Nvidia L4	datacenter	24GB	$0.81/hr	$0.81/hr–$0.81/hr	1	Compare →
Nvidia A10	datacenter	24GB	$1.01/hr	$1.01/hr–$1.01/hr	1	Compare →
Nvidia Tesla V100	datacenter	16GB	$3.06/hr	$3.06/hr–$3.06/hr	1	Compare →
Nvidia A100	datacenter	80GB	$3.20/hr	$3.20/hr–$3.20/hr	1	Compare →
Nvidia Nvidia A100 80GB PCIe	datacenter	40GB	$4.10/hr	$4.10/hr–$4.10/hr	1	Compare →
Nvidia H100	datacenter	80GB	$12.29/hr	$12.29/hr–$12.29/hr	1	Compare →
Nvidia B200	datacenter	192GB	$20.00/hr	$20.00/hr–$20.00/hr	1	Compare →

Compare with peers

Other providers in the same bucket — quick way to sanity-check pricing before committing.

Google Cloud

Cheapest: $0.35/hr · Nvidia T4 · 7 GPUs

Compute Engine A2/A3 (A100, H100), G2 (L4), and TPU instances. First-class integration with Vertex AI for managed training and serving. Sustained-use discoun...

Microsoft Azure

Cheapest: $0.53/hr · Nvidia T4 · 6 GPUs

ND/NC/NV-series VMs covering A100, H100, H200, MI300X. Tight integration with Azure ML and OpenAI Service. Reserved Instances and Spot VMs offer up to 90% of...

Oracle Cloud

Cheapest: $2.00/hr · Nvidia A10 · 5 GPUs

Bare-metal and VM GPU shapes (A10, A100, H100, H200) with RDMA cluster networking that often beats other hyperscalers on $/GPU-hour for large training jobs.

V Vultr

Cheapest: $1.45/hr · Nvidia NVIDIA A40 · 5 GPUs

Cloud GPU + Bare Metal GPU instances. Globally distributed datacenters (32 regions), per-hour billing, transparent published pricing. Sweet spot for teams wa...

Frequently asked

How is Amazon Web Services billing measured?

Most providers bill per-second once the instance is running, with a small minimum (often 60 seconds). Some first-party clouds round up to the minute. Headline $/hr is the right comparison unit.

Which regions does Amazon Web Services offer GPUs in?

P2P marketplaces aggregate hosts worldwide so region varies per offer. First-party clouds and hyperscalers expose explicit region pickers (US, EU, APAC). Filter on the provider's site after clicking through if region matters.

Does Amazon Web Services offer an SLA?

Hyperscalers (AWS, GCP, Azure, Oracle) publish formal SLAs. First-party clouds (Lambda, CoreWeave) offer support contracts. P2P marketplaces and decentralized networks have no SLA — uptime depends on the individual host.

What's the refund / cancellation policy on Amazon Web Services?

Per-second billing means you only pay for compute used — stop the instance and billing stops. Pre-paid credits and committed-use discounts have provider-specific terms; check the provider's billing docs before pre-paying.