01GPU cloud· manual survey 2026-07-06 · automated rows 2026-07-23
What compute rents for
A venue-by-venue view of accelerator pricing with billing and evidence kept intact. Pick a chip — the statistics, history, and quotes below all follow it.
H100 · primary on-demand floor
$1.99/GPU·hr
Voltage Park · specialist cloud
H100 · primary on-demand median
$3.95/GPU·hr
10 comparable provider-list quotes
H100 · on-demand hyperscaler floor
$6.88/GPU·hr
AWS (p5) · On-demand
H100 · venue spread
- hyperscaler floor ÷ marketplace floor
- 4.0× AWS (p5) $6.88/GPU·hr ÷ Vast.ai $1.73/GPU·hr
- specialist median ÷ marketplace floor
- 2.2× median $3.88/GPU·hr (Voltage Park $1.99 · Hyperstack $3.20 · RunPod $3.29 · Nebius $3.85 · Crusoe $3.90 · Lambda $3.99 · Together AI $3.99 · CoreWeave $6.16) ÷ Vast.ai $1.73/GPU·hr
Different venues sell different things — SLAs, egress, and commitment terms — so the gap measures venue dispersion, not an arbitrage.
Comparable price history
Loading evidence ledger…
Provider dispersion
H100 · comparable provider-list range
Provider-primary on-demand quotes under primary-on-demand-v1. The implied anchor is a non-executable market reference, not a current provider quote.
- Azure (ND v5)On-demand$12.29 per GPU hour; Hyperscaler; On-demand; observed 2026-07-23
- AWS (p5)On-demand$6.88 per GPU hour; Hyperscaler; On-demand; observed 2026-07-23
- CoreWeaveOn-demand$6.16 per GPU hour; Specialist cloud; On-demand; observed 2026-07-06
- LambdaOn-demand$3.99 per GPU hour; Specialist cloud; On-demand; observed 2026-07-06
- Together AIOn-demand$3.99 per GPU hour; Specialist cloud; On-demand; observed 2026-07-06
- CrusoeOn-demand$3.90 per GPU hour; Specialist cloud; On-demand; observed 2026-07-06
- NebiusOn-demand$3.85 per GPU hour; Specialist cloud; On-demand; observed 2026-07-06
- RunPodOn-demand$3.29 per GPU hour; Specialist cloud; On-demand; observed 2026-07-06
- HyperstackOn-demand$3.20 per GPU hour; Specialist cloud; On-demand; observed 2026-07-06
- Voltage ParkOn-demand$1.99 per GPU hour; Specialist cloud; On-demand; observed 2026-07-06
Current published quotes
Loading the sortable table and expandable mobile evidence cards…
Anchor board · implied median vs venue quotes
Weekly exact-time implied medians for B200/H200/H100/A100 and the RTX 5090 front-month monthly-average median (Kalshi, AUG 28 2026 · 22:35 UTC), beside primary on-demand provider lists and separately labeled marketplace access asks (manual survey 2026-07-06; automated rows through 2026-07-23). Capacity blocks, serverless, and secondary estimates are excluded. Anchors are not current spot quotes.
| Chip | Implied anchor | Prior snapshot | 30D same contract | Front-month avg | Hyperscaler | Specialist | Marketplace |
|---|---|---|---|---|---|---|---|
| B200 price historyHBM3e · 180GB | $6.03Sep 4 weekly | New contractno comparable prior | Building history | $6.25 | $14.241 | $5.89–$8.602 | $6.363 |
| H200 price historyHBM3e · 141GB | $4.59Sep 4 weekly | New contractno comparable prior | Building history | $4.75 | $7.91–$10.604 | $3.99–$6.315 | $3.936 |
| H100 price historySXM · 80GB | $3.00Sep 4 weekly | New contractno comparable prior | Building history | $2.86 | $6.88–$12.297 | $1.99–$6.168 | $1.73–$2.049 |
| A100 price historySXM4 · 80GB | $1.14Sep 4 weekly | New contractno comparable prior | Building history | $1.13 | — | $1.35–$2.7910 | — |
| RTX 5090 price historyGDDR7 · 32GB | $0.62Aug ’26 monthly avg | +6.9%6d 3h 35m | +12.7%3 closes · 08-21–08-28 | $0.62 | — | $0.9911 | $0.3112 |
Provider quote sources12 notes
- 1 AWS (p6) $14.24
- 2 RunPod $5.89 · Hyperstack $6.00 · Lambda $6.69 · Nebius $7.15 · Together AI $8.19 · CoreWeave $8.60
- 3 Vast.ai $6.36
- 4 AWS (p5en) $7.91 · Azure (ND v5) $10.60
- 5 Hyperstack $3.99 · Crusoe $4.29 · RunPod $4.39 · Nebius $4.50 · CoreWeave $6.31
- 6 Vast.ai $3.93
- 7 AWS (p5) $6.88 · Azure (ND v5) $12.29
- 8 Voltage Park $1.99 · Hyperstack $3.20 · RunPod $3.29 · Nebius $3.85 · Crusoe $3.90 · Lambda $3.99 · Together AI $3.99 · CoreWeave $6.16
- 9 Vast.ai $1.73 · SF Compute $2.04
- 10 Hyperstack $1.35 · RunPod $1.49 · Crusoe $2.30 · CoreWeave $2.70 · Lambda $2.79
- 11 RunPod $0.99
- 12 Vast.ai $0.31
Marketplace tape · intraday ask floor — listed offers, not clearing prices
Vast.ai verified on-demand asks per chip, captured on an hourly schedule and shown over the last 48h. 133 hourly captures across 11 days, through 2026-09-01 10:31 UTC. Missed hours are gaps, never interpolated.
- H100$1.73floormedian $2.58 · 18 offers · 10/48h observed
- H200$3.98floormedian $4.76 · 11 offers · 11/48h observed
- B200$5.31floormedian $6.00 · 25 offers · 11/48h observed
- RTX 4090$0.30floormedian $0.40 · 61 offers · 11/48h observed
- RTX 5090$0.33floormedian $0.47 · 64 offers · 11/48h observed
Ask floor — listed offers, not clearing prices. Full rolled archive: /marketplace-tape.json.