Skip to main content

AI Compute Price (H100)

Tracks the price of AI compute, anchored to NVIDIA H100.

This market represents the market price of the NVIDIA H100, the dominant high-end GPU used to train and run modern large-scale AI models. As the primary hardware constraint for AI deployment, H100 pricing reflects real demand for compute across data centers, model training, and infrastructure buildouts.

Specifically, the market tracks the price of the NVIDIA H100 SXM5 80GB HBM3 — the data center variant designed for multi-GPU training and high-throughput inference workloads.

Data Sources​

We fetch source data from the following providers:

  • Vast.ai - Distributed cloud computing marketplace for GPU rental services
  • RunPod - Cloud platform specializing in GPU compute for AI and machine learning
  • Lambda Labs - GPU cloud provider focused on deep learning and AI workloads
  • Market mid price - Current market's mid price from Tings market

All sources are weighted equally in our oracle calculation. These coefficients might be adjusted during the testnet phase as we assess the reliability and accuracy of each source.

Oracle Algorithm​

The final oracle value is calculated using the oracle algorithm described in the Oracle Price Evaluation documentation.