Back
NVIDIA H200 Pricing (August 2025): Cheapest On‑Demand Cloud GPU Rates
Who really has the lowest H200 price? We scraped every public U‑S region price list so you don’t have to.
Published:
Aug 1, 2025
Last updated:
Aug 1, 2025
One‑minute snapshot
Provider | SKU / Instance | On‑Demand $/GPU‑hr* | Notes |
|---|---|---|---|
AWS (p5e.48xlarge) | 8× H200 141 GB | $4.33 (1-day minimum) | Capacity‑Blocks pricing is minimum of 1 day; divide $34.608 by 8 (Amazon Web Services, Inc.) |
Azure (Standard ND96isr H200 v5) | 8× H200 | $10.60 | Calculator price $84.8 /hr total (Public Cloud Reference) |
Google Cloud | A3 H200 (on‑demand) | TBA | Google lists H200 only as Spot for now; on‑demand not yet published (Jarvislabs.ai Docs) |
Oracle Cloud (BM.GPU.H200.8) | 8× H200 | $10.00 | Bare‑metal node, $80 /hr total (Oracle) |
Lambda Cloud (HGX H200) | 1× H200 | $3.79 | Minute‑billed, no commitment (Lambda) |
CoreWeave (8 × H200) | 8× H200 | $6.31 | $50.44 /hr node / 8 GPUs (ionstream.ai) |
RunPod (8 × H200) | 8× H200 | $3.99 | $31.92 /hr node / 8 GPUs (ionstream.ai) |
1× H200 | $3.80 | Single‑GPU VM, pay‑as‑you‑go (Jarvislabs.ai Docs) | |
Marketplace | ≈ $2.43 | Lowest current host listing (GetDeploying) |
*Prices are normalized per single H200 even when a provider sells only 8‑GPU nodes. U‑S region pricing, on‑demand only (no spot, reserved, or contract rates).
Methodology – why you can trust these numbers
On‑demand only. We excluded capacity reservations longer than 14 days, reserved instances, and spot/pre‑emptible offers.
Same silicon. Every row is a 141 GB NVIDIA H200 (SXM or PCIe).
Public price lists only. Figures come straight from each provider’s pricing page on 25 July 2025.
US regions, USD. Regional variation can add 5‑20 percent; those are ignored for apples‑to‑apples comparison.
A100 vs H200 cost‑performance benchmark
Provider | 2 hrs runtime | Effective cost |
|---|---|---|
Thunder Compute – A100 80 GB | 2 × $0.78 | $1.56 |
Vast.ai – H200 | 2 × $2.43 | $4.86 |
RunPod – H200 | 2 × $3.99 | $7.98 |
Lambda – H200 | 2 × $3.79 | $7.58 |
AWS – H200 | 2 × $4.33 | $8.66 |
Azure – H200 | 2 × $10.60 | $21.20 |
Bottom line: two hours on Thunder Compute’s A100 costs less than 15 minutes on Azure’s H200—and still buys roughly 13 × more runtime per dollar than hyperscaler H200s.
Takeaways for developers
H200 premiums remain steep. Even after AWS’s June price cut, the cheapest hyperscaler H200 hour is >5 × Thunder Compute’s A100.
Specialist clouds narrow the gap. Lambda, RunPod, Jarvislabs, and Vast.ai all sit in the $2–4 range, but Thunder’s A100 is still ~2‑5 × cheaper.
Choose H200 only when you must—massive models that overflow 80 GB VRAM or long‑context inference. For prototyping, fine‑tuning, and most training, A100 80 GB wins on ROI.
Thunder Compute roadmap. We don’t offer H200 nodes yet; today you can launch A100 80 GB at $0.78/hr (one‑click VS Code, per‑second billing, persistent volumes, live hardware swaps).
Bookmark this page—we refresh the numbers quarterly. Meanwhile, spin up an A100 or H100 on Thunder Compute and keep more budget for your model weights.
Carl Peterson
Other articles you might like
Learn more about how Thunder Compute will virtualize all GPUs
Try Thunder Compute
Start building AI/ML with the world's cheapest GPUs