H100 GPU Rental Cost Per Hour 2026: All Providers

H100 GPU rental cost per hour 2026: cheapest verified rate is Vast.ai interruptible at $1.49/GPU-hr (July 2026); RunPod Community Cloud H100 PCIe at $1.99/GPU-hr; specialist-cloud H100 SXM at $2.99-$6.16/GPU-hr; hyperscaler on-demand at $6.88-$12.29/GPU-hr. Reserved-capacity contracts unlock 30-60% off list for predictable workloads.

Last verified: Sep 16, 2026.

At a glance

• Vast.ai marketplace floor: $1.49/GPU-hr interruptible (cheapest published)
• RunPod Community Cloud H100 PCIe: $1.99/GPU-hr; SXM $2.69/GPU-hr
• RunPod Secure Cloud H100 PCIe: $2.89/GPU-hr; SXM $3.29/GPU-hr
• Lambda 1x H100 PCIe: $3.29/GPU-hr; H100 SXM $3.99/GPU-hr
• CoreWeave HGX H100 (8x): $6.16/GPU-hr on-demand, $2.46/GPU-hr spot
• AWS Capacity Blocks p5.4xlarge: $5.191/GPU-hr effective reserved
• Azure ND H100 v5: $12.29/GPU-hr on-demand (highest)
• GCP A3 a3-highgpu-8g: $11.06/GPU-hr on-demand, $3.69/GPU-hr spot

Why H100 rental rates vary 8x for identical silicon

A verified H100 SXM ranges from $2.99/GPU-hr to $12.29/GPU-hr in 2026 — the spread comes from what surrounds the GPU.

Five structural variables move the rate: (1) instance size — 1x vs 8x node; (2) chip form factor — PCIe vs SXM (SXM runs NVLink at 900 GB/s, required for multi-GPU gradient synchronization); (3) included CPU/RAM/storage — bare accelerator vs fully provisioned VM; (4) billing commitment — interruptible spot vs on-demand vs reserved 1-3 year; (5) provider tier — marketplace host, neocloud, hyperscaler (answerfootprint.com, September 2026).

The cheapest rates come from hosts with the least overhead. Vast.ai's marketplace aggregates independent GPU owners willing to rent at $1.49/GPU-hr interruptible — the cost of interruptibility and host-level reliability variance. RunPod Community Cloud at $1.99/GPU-hr for H100 PCIe is next-cheapest because RunPod does not own the silicon; third-party hosts supply it. The most expensive rate — Azure ND H100 v5 at $12.29/GPU-hr — pays for compliance certifications, multi-region availability, and a 4-week capacity-quota approval process for 8+ node requests (cloudzero.com, September 2026).

Specialist-cloud H100 pricing

Specialist clouds are the practical middle layer between marketplace rates and hyperscaler list.

Lambda Labs runs the most transparent published card with verified August 2026 rates: H100 PCIe 1x at $3.29/GPU-hr; H100 SXM 1x at $4.29/GPU-hr; H100 SXM 8x (208 vCPUs, 1.8 TB RAM, 22 TB SSD) at $3.99/GPU-hr (lambda.ai/pricing). Lambda Reserved Cloud 1-Click Cluster pricing drops further: 16-GPU HGX H100 at $6.16/GPU-hr; 64-GPU at $5.85; 256-GPU at $5.54 — all on 2-week to 1-year terms. Lambda's H100 capacity is intermittent: multi-week waitlists during demand spikes are the recurring complaint. The recurring complaint in 2026 is capacity, not price.

RunPod offers two tiers with the same hardware: Community Cloud (third-party hosts) at $1.99/GPU-hr H100 PCIe / $2.69/GPU-hr H100 SXM / $3.59 H200; Secure Cloud (RunPod's own datacenters) at $2.89/GPU-hr H100 PCIe / $3.29/GPU-hr H100 SXM / $4.59 H200. RunPod bills per-second (1-minute minimum). Storage is metered separately. The Community-to-Secure spread is the cost of platform-level reliability vs host-level reliability.

CoreWeave publishes an 8-GPU-node rate card, not per-GPU rates. HGX H100 8x at $49.24/hr on-demand = $6.16/GPU-hr; spot $19.71/hr = $2.46/GPU-hr (40-60% off on-demand). HGX H200 8x at $50.44/hr = $6.31/GPU-hr; spot $20.93/hr = $2.62/GPU-hr. The 8-GPU minimum is the real cost — single H100 is not bookable on CoreWeave at any price except GH200 (1x at $6.50/hr). Reserved contracts up to 60% off, all sales-negotiated. CoreWeave's business model runs on the third tier (OpenAI $22.4B cumulative, Meta $21B through 2032); the rate card is for everyone who hasn't negotiated yet (cloudzero.com, September 2026).

Hyperscaler H100 pricing

At 8x node level, AWS, GCP, and Azure cluster within 15 cents of each other at roughly $98.32-$98.46/hr — but per-GPU normalization exposes a 4x spread.

AWS p5.48xlarge (8x H100 80GB, 640 GB total HBM3) lists at $55.04/hr on-demand, $41.528/hr effective under Capacity Blocks for ML (US East), $23.777/hr on 1-year Reserved, $19.769/hr on spot (vantage.sh, July 2026). Capacity Blocks require committing to a future time window — middle ground between on-demand and reserved. A separately-priced aggregator lists a single-GPU p5.4xlarge on-demand at $6.88/GPU-hr. AWS announced an 'up to 45%' price reduction on H100 in June 2025; the current effective rate is the result.

Google Cloud a3-highgpu-8g (8x H100) lists at $88.49/hr on-demand, drops to $38.86/hr on 3-year committed use discount (CUD) — a 55% discount versus list (gpusmith.com, July 2026). The newer a3-megagpu-8g with enhanced GPU-to-GPU networking runs $93.40/hr on-demand, $40.65/hr on 3-year CUD. GCP's a3-ultragpu-8g ships with 8x H200 (not H100) at $84.81/hr on-demand — confusing naming, worth verifying SKU. Spot pricing on GCP H100 runs $2.10-$2.80/GPU-hr depending on region.

Azure ND H100 v5 series (Standard_ND96isr_H100_v5, 8x H100, 96 vCPUs, 1.9 TB RAM) is the most expensive at $98.32/hr on-demand ($12.29/GPU-hr normalized) — consistent across July 2026 pricing surveys (vantage.sh, spheron.network). 1-year reservation brings it to ~$63.40/hr ($7.93/GPU-hr); 3-year reservation to ~$43.79/hr ($5.47/GPU-hr). Azure's quota approval process for 8+ node requests can stretch to 3-4 weeks before capacity is confirmed — a structural disadvantage for time-sensitive training jobs.

Vast.ai and the marketplace floor

Vast.ai's marketplace delivers the lowest published rates in the category at the cost of interruptibility and host variance.

Verified H100 PCIe prices on Vast.ai: unverified community hosts at $0.90-$1.60/GPU-hr; datacenter-verified hosts at $1.50-$2.27/GPU-hr (spheron.network, July 2026). The Vast.ai 'on-demand' label does not mean what it means on AWS or GCP — it means the host is currently offering the machine at the listed price and will not actively evict you, but if the host takes the hardware offline for any reason, the machine stops. There is no platform-level guarantee of availability. For fault-tolerant training with checkpoint discipline, Vast.ai interruptible at $1.49/GPU-hr is the cheapest credible option in 2026.

Three rules for Vast.ai: (1) sort by DLPerf score for reliability; (2) only use datacenter-verified hosts for production jobs; (3) always checkpoint every 30 minutes or less because interruptible means reclaimable without notice. The platform does charge for storage while the instance is paused, which adds 5-10% to total cost for low-utilization workflows.

Reserved-capacity economics

For predictable training workloads, reserved capacity delivers 30-60% off list — but the headline rates are misleading.

Together AI's reserved cluster (91-180 day) starts at $3.09/GPU-hr for H100 SXM. Lambda 1-Click Cluster 16-GPU reservation drops to $6.16/GPU-hr on 2-week-1-year terms, $5.54/GPU-hr at 256-GPU scale. CoreWeave reserved capacity ceiling is 'up to 60% off on-demand' — target $2.46/GPU-hr reserved for H100 — but only achievable through sales negotiation. Hyperscaler 1-year reservations bring H100 to $7-$8/GPU-hr; 3-year reservations to $5-$6/GPU-hr (nOps FinOps, August 2026).

The break-even calculation: for a job that runs continuously for 6+ months, reserved beats on-demand even with the deposit. For a workload that might run for 2 weeks and then stop, on-demand wins because reserved capacity is wasted if unused. The reservation deposit on hyperscalers is typically 30-50% of the 1-year value, refundable only if you commit to a longer term. Most mid-market teams should avoid 3-year reservations unless they have locked-in demand.

Side-by-side H100 pricing comparison

How to choose the right H100 provider in 2026

The decision hinges on three variables: workload duration, reliability tolerance, and budget.

1. For a 1-day fine-tuning experiment: Vast.ai verified datacenter at $1.49/GPU-hr. Accept interruptibility; checkpoint every 30 minutes.
2. For a 1-2 week iterative fine-tune: RunPod Community Cloud H100 SXM at $2.69/GPU-hr or Lambda 1x H100 PCIe at $3.29/GPU-hr. Per-second billing lets you stop and restart cheaply.
3. For a 1-3 month training run you can't afford to lose: Lambda Reserved Cloud 16-GPU cluster at $6.16/GPU-hr, or Together AI reserved at $3.09/GPU-hr with commitment.
4. For a 200+ GPU distributed training job: CoreWeave HGX H100 with reserved negotiation. The 8-GPU node minimum and InfiniBand interconnect pay for themselves at scale; naive scaling on a cheaper cluster runs at 70% efficiency and costs more per useful FLOP.
5. For production inference with compliance requirements: AWS Capacity Blocks at $5.191/GPU-hr, or AWS 3-year reserved at $5-$6/GPU-hr. The integration with SageMaker and enterprise support justify the premium.

What enterprise buyers should do next

Three actions for organizations evaluating H100 capacity in 2026.

1. Run a 2-week pilot across 2-3 providers on the same workload. Same training script, same dataset, same hyperparameters. Compare cost, throughput, and reliability. The cheapest published rate is rarely the cheapest end-to-end cost once you factor in support responsiveness and reclaim frequency.
2. Negotiate a reserved contract before committing to a single provider. CoreWeave, Lambda, and Together AI all negotiate reserved rates meaningfully off list. Even a 1-year commitment typically unlocks 25-40% off list. The published rate card is a starting point, not the final price.
3. Model egress and storage costs separately. Hyperscalers charge $0.087-$0.12/GB egress; Lambda, CoreWeave, and Crusoe publish free egress. For a 10 TB dataset with monthly retraining, egress alone adds $870-$1,200/month on AWS — more than the savings from a lower H100 hourly rate.

What to watch next

Three near-term datapoints. First, the CoreWeave Q3 2026 earnings call — CoreWeave CFO stated the company is 'largely sold out' of 2026 capacity and raising list prices; expect rate-card increases by Q4 2026. Second, AWS's planned p6-b200.48xlarge at $98.84/hr effective — Blackwell-generation capacity will relieve H100 demand pressure and likely pull H100 list rates down 10-15% by mid-2027. Third, NVIDIA Vera Rubin full production at CoreWeave, Google, Azure, OCI, and Nebius — Rubin GPUs hit $40B/gigawatt revenue per NVIDIA's earnings, and as Rubin capacity comes online, H100 spot prices will fall faster than reserved rates (NVIDIA Q2 FY27 earnings, August 26, 2026).