CT-MKT
AI Compute Market Rate
- Value
- $6.60 / market H100e-hr
- Daily move
- n/a
- Status
- Stable
- Signal
- Daily observed H100-equivalent market rate across qualified hyperscaler, AI-native cloud, and developer GPU cloud capacity.
Market WatchIndex Status
Market
Market
AI compute pricing, capacity, and availability signals.
Track H100-equivalent cloud pricing, spot compute, training economics, inference cost, and GPU availability across the AI infrastructure market. Enterprise pricing stable · Observed market rate stable · Spot market easing · Availability slightly improved.
6 measuresIndex family
CT-MKT tracks observed market movement.
CT-SPOT tracks interruptible compute pricing.
CT-GPUA measures practical GPU availability.
CT-COST tracks enterprise AI cloud pricing.
CT-TRAIN converts pricing into standard training cost.
CT-INFER converts pricing into standard inference cost.
Current Snapshot
Current ComputeTape Index Snapshot
Live value cards with daily move and signal text. Last updated May 17, 2026, 6:00 AM EDT.
CT-SPOT
AI Spot Compute Index
- Value
- $2.73 / spot H100e-hr
- Daily move
- -2.6%
- Status
- Looser
- Signal
- Interruptible and spot H100-equivalent compute pricing.
CT-GPUA
GPU Availability Index
- Value
- 70.0 / 100 availability
- Daily move
- 0 pts
- Status
- Slightly improved
- Signal
- Practical AI GPU availability on a 0-100 scale.
CT-COST
Enterprise AI Cloud Cost Index
- Value
- $8.41 / H100e-hr
- Daily move
- 0.0%
- Status
- Stable
- Signal
- Realistic enterprise-grade H100-equivalent AI cloud pricing across major hyperscaler and AI-native cloud providers.
CT-TRAIN
AI Training Cost Index
- Value
- $70,866 / training run
- Daily move
- 0.0%
- Status
- Stable
- Signal
- Estimated infrastructure cost to train a standard 7B dense model on 1 trillion tokens.
CT-INFER
AI Inference Cost Index
- Value
- $3.40 / 1M tokens
- Daily move
- 0.0%
- Status
- Stable
- Signal
- Estimated infrastructure cost to generate 1 million output tokens from a standard 70B-class inference workload.
Pressure
Market Pressure
A daily read on whether enterprise AI compute markets are tightening, easing, or staying stable. Enterprise AI compute markets remain stable overall. Headline enterprise pricing is unchanged, and the observed market rate is stable in the CT-MKT v1 basket. Spot capacity softened slightly. GPU availability improved modestly, though reliable large-cluster access remains constrained. Updated daily · Last updated May 16, 2026. Power conditions are a key driver of compute availability.
StableEnterprise pricing
Headline enterprise rates are unchanged.
StableObserved market rate
CT-MKT daily move: Daily n/a.
EasingSpot market
Interruptible capacity improved across tracked providers.
Slightly improvedGPU availability
Access improved, but reliable large clusters remain constrained.
Providers
Provider Price Board
Snapshot values for normalized H100-equivalent pricing across tracked AI cloud providers. Prices distinguish enterprise on-demand, qualified market, secure cloud, spot, and capacity-style products.
| Provider | Segment | GPU | Mode | Region | Raw $ / GPU-hr | H100e $ / hr | Adjusted H100e $ / hr | Availability | Confidence | Last checked | Source |
|---|---|---|---|---|---|---|---|---|---|---|---|
| AWS | Hyperscaler | H100 | on-demand | US | $6.88 | $6.88 | $7.22 | Quota Gated | High | May 16, 2026, 4:00 AM EDT | |
| Azure | Hyperscaler | H100 | on-demand | US | $12.29 | $12.29 | $12.90 | Quota Gated | High | May 16, 2026, 4:00 AM EDT | |
| Oracle | Hyperscaler | H200 | on-demand | US | $10.00 | $8.70 | $8.96 | Available Limited | High | May 16, 2026, 4:00 AM EDT | |
| CoreWeave | Ai Native Cloud | H100 | on-demand | US | $6.16 | $6.16 | $6.34 | Available Limited | High | May 16, 2026, 4:00 AM EDT | |
| Crusoe | Ai Native Cloud | H100 | on-demand | US | $5.80 | $5.80 | $5.97 | Available Limited | Medium | May 16, 2026, 4:00 AM EDT | |
| Nebius | Ai Native Cloud | H100 | cloud | EU/US | $5.25 | $5.25 | $5.41 | Available Limited | Medium | May 16, 2026, 4:00 AM EDT | |
| Lambda | Developer Gpu Cloud | H100 | on-demand | US | $4.25 | $4.25 | $4.25 | Available | Medium | May 16, 2026, 4:00 AM EDT | |
| RunPod Secure | Developer Gpu Cloud | H100 | secure-cloud | US | $2.99 | $2.99 | $3.08 | Available Limited | Medium | May 16, 2026, 4:00 AM EDT | |
| Qualified Marketplace Median | Qualified Marketplace | H100 | qualified-marketplace | US/EU | $3.80 | $3.80 | $3.91 | Available Limited | Medium | May 16, 2026, 4:00 AM EDT |
Calculation
CT-MKT Calculation Notes
CT-MKT uses segment medians across hyperscalers, AI-native clouds, developer GPU clouds, and qualified marketplace observations. Prices are normalized into H100-equivalent GPU-hours and adjusted for availability friction. MVP observations are research estimates until source polling is automated.
40 / 35 / 20 / 5Segment basket
Hyperscalers 40%, AI-native clouds 35%, developer GPU clouds 20%, qualified marketplace 5%.
9Included observations
Mixed / Medium-High confidence. Low-confidence and unavailable observations are excluded.
ct-mkt-v1Calculation version
Last observed May 16, 2026, 4:00 AM EDT.
Access
Capacity Board
Availability and access signals for major AI compute providers.
| Provider | H100 | H200 | B200 / GB200 | MI300X | Large Cluster | Quota Friction | Status |
|---|---|---|---|---|---|---|---|
| AWS | Yes | Yes | Emerging | No | Strong | High | Constrained |
| Azure | Yes | Yes | Emerging | Some | Strong | High | Constrained |
| Oracle | Yes | Yes | Emerging | No | Strong | Medium | Limited |
| CoreWeave | Yes | Yes | Emerging | No | Strong | Medium | Available / tight |
| Lambda | Yes | Some | No | No | Moderate | Low/Med | Available |
| RunPod | Yes | Some | No | Some | Lower | Low | Variable |
| Crusoe | Yes | Some | Emerging | No | Strong | Medium | Expanding |
| Nebius | Yes | Some | Emerging | No | Strong | Medium | Expanding |
Interruptible
Spot Market Board
Interruptible and spot GPU prices can move faster than enterprise list prices. These signals inform CT-SPOT and market pressure, not CT-COST.
| Provider | GPU | Spot / Interruptible $ | Daily Move | Availability | Notes |
|---|---|---|---|---|---|
| CoreWeave | H100 | $2.46 | -1.2% | Moderate | Spot supply improved |
| RunPod | H100 | $2.40-$3.10 | -2.0% | Variable | Marketplace depth varies |
| Vast | H100 | $1.80-$2.80 | -3.5% | Variable | Quality varies |
| AWS | H100 | Unavailable | Flat | Region-specific | Enterprise-grade spot |
| Lambda | H100 | Unavailable | Flat | Limited | Track if available |
Spot and marketplace capacity may be interruptible, region-limited, or unsuitable for large training clusters.