The Risk-Transfer
Layer for the AI Economy
GPU compute is the most important commodity of the decade. We're building the first financial products to hedge its price — so you can lock in costs and let someone else take the volatility.
GPU compute has no
financial infrastructure.
Compute is becoming a traded input factor like power, bandwidth, or FX. Yet the market operates like oil did a century ago: no benchmark pricing, no hedging instruments, no way to manage risk.
Downward Drift
Technology progress drives long-term cost reduction. H100 prices dropped 60%+ in 18 months.
Sudden Upward Jumps
Supply shocks and policy changes cause surprise hikes. AWS raised H200 prices 15% overnight.
Massive Dispersion
H200 pricing varies $3.7–$10.6/hr across providers — a 3x spread based on region and vendor.
Drivers of Volatility
Supply & Demand Shocks
Limited availability of advanced GPUs pushes prices up unexpectedly. H100/H200 shortages caused sudden cost spikes across all providers.
Cloud Economics Shifts
Cloud providers increasingly treat GPU capacity as scarce infrastructure — variable pricing tied to booking demand and portability.
Market Competition
Specialist providers undercut hyperscalers, creating wide spreads and price wars. H200 pricing varies nearly 3x across providers.
Compute Inflation Swaps.
Risk transfer, not speculation.
A fixed-for-floating swap where the floating leg is tied to a compute price index. Cash-settled. No physical delivery. No vendor lock-in. Exactly like an interest rate or commodity swap.
How the Swap Works
Cash-Settled
No physical delivery of compute. No dependency on vendor uptime, performance, or usage verification. Settlement is purely financial.
Hedge, Not Commitment
Unlike cloud commitments that fix volume, our swap fixes price risk. Preserves vendor flexibility and avoids over-commitment if demand falls.
CFO-Ready
Qualifies as a cash-flow hedge. Auditable cashflows. Clean disclosure. Treated as hedge, not capex. No operational dependency on cloud usage.
Sample Terms
The CGPI.
The reference price for compute.
The Compute GPU-Hour Price Index tracks volume-weighted average spot prices across major hardware types. It's the foundation for pricing, hedging, and settlement.
Built on real trades. Not scraped offers.
The CGPI is a volume-weighted average of eligible GPU compute transactions, normalized to H100-equivalent performance basis. Conservative, transparent, boringly reliable.
- GPU cloud providers
- Secondary marketplaces
- On-demand pricing
- Bilateral trade prints
- Top/bottom 10% excluded
- No source > 30% weight
- Monthly final, weekly indicative
- 30-day change notice
Financial products built
for compute exposure.
Cash-settled contracts that let you lock in prices, hedge revenue, or take a position on the most important commodity of the decade.
For Compute Buyers
Hedge inflation riskLock in predictable compute costs for upcoming models and product launches. Protect your P&L from GPU price shocks without vendor lock-in.
For Compute Providers
Hedge deflation riskStabilize cash flows so you can finance new build-outs with certainty. Convert volatile spot upside into predictable yield.
For Lenders
De-risk portfoliosReduce exposure to the underlying compute price risk of the GPU assets you finance. A balance-sheet hedge for your portfolio.
For Traders
New asset classTake directional views on the future of compute as a commodity. A new exposure orthogonal to traditional asset classes.
Transparent.
Institutional-Grade.
Designed for CFOs first. Traders adapt automatically. Every design decision optimizes for trust, simplicity, and auditability.
Transaction-Based Pricing
Our index is built on real trades. Not scraped offers, not surveys, not estimates. Volume-weighted from a diversified set of compute providers.
Manipulation-Resistant
Hard caps on contributor weights, outlier rejection, automatic fallback rules. No single participant can economically move the index.
Conservative Governance
Methodology changes require 30-day notice. Historical values are never restated. Emergency adjustments only for data unavailability.
Compliant Architecture
A non-regulatory financial benchmark designed for settlement and risk management. ISDA definitions incorporated by reference. New York law.
OTC first.
Then the exchange.
Bilateral Hedges
- Publish GPU-hour index
- Structure 5–10 bilateral swaps
- Learn where pricing breaks
Standardization
- Introduce caps & corridors
- Standardize tenors
- Onboard financial counterparties
Exchange
- Clearing infrastructure
- Performance-adjusted indices
- Token-native derivatives
Compute is the Next Commodity Exchange
The financial layer
for AI infrastructure.
The compute economy is here. ForwardCompute is its risk-transfer layer. Talk to our team about structuring your first hedge.