Two-Settlement
Compute
Marketplace
Day-ahead scheduling, real-time balancing, system-level resource allocation.The market rails that AI deserves.
Data center capex by 2030
Average GPU utilization
Globally optimal markets — for now
The ultimate resource
allocation problem
AI compute allocation relies on long-term bilateral contracts and static provisioning. The result is persistent, structural inefficiency at exactly the moment demand is exploding.
Chronic Underutilization
Inference workloads spike. Training jobs batch. Operators overprovision for peak demand, and billions in stranded capacity sit idle.
Tough Provider Economics
Independent datacenters accept wholesale offtakes to secure stable cashflow, compressing margins across every new entrant.
No Price Discovery
Prices are negotiated privately in bilateral contracts. No mechanism exists to discover the real-time value of compute.
Volatility Without Hedging
Demand swings wildly while prices stay fixed. No forward market, no real-time balancing, no way to hedge exposure.
Rebuild efficiency from the ground up
A two-settlement market for primary and secondary clearing — the structure that made electricity markets efficient, applied to AI compute.
Predictable Scheduling
Schedule workloads 24 hours ahead. Lock capacity commitments. Clear at auction-determined prices that reflect expected scarcity.
- Hedge real-time exposure
- Hardware & latency constraints
Live Dispatch
Handle deviations as they occur. Dispatch flexible capacity to meet realized demand. Prices form continuously on marginal cost.
- Sub-second dispatch
- Inference spike handling
A standardized unit of compute that makes heterogeneous hardware comparable. Performance factors derived from MLCommons benchmarks translate raw capacity into tradeable units.
Three sides of one market
Compute Buyers
AI Labs · Enterprises
Access capacity on demand without long-term commitments. Hedge training costs and scale inference elastically.
Compute Providers
Hyperscalers · Neo-Clouds
Monetize idle capacity through transparent markets. Reduce overprovisioning with network-level pooling.
Liquidity Providers
Traders · Market Makers
Earn availability payments for standby capacity. Capture spreads between day-ahead and real-time markets.
Volatility is a local phenomenon
When workloads pool across the network, individual spikes become statistical noise. The law of large numbers absorbs volatility at scale.
Reduction
Efficiency
Kill the intermediary spread
EIGEN clears directly between providers and end users, capturing value currently lost to wholesale offtake structures.
Providers accept low wholesale rates for stability; users pay retail markups.
Direct clearing cuts user cost 23% while tripling provider margin.
Make compute efficient again.
It’s ambitious, we know. But it’s too important not to try. We’re looking for partners across the compute ecosystem to rebuild the market rails of intelligence.
jasonsun@stanford.edu