OTC forwards, swaps, and structured products for the GPU compute market. Bringing transparent price discovery and hedging to AI infrastructure.
GPU compute is the most important commodity of the AI era — yet it trades like crude oil before NYMEX. Bilateral contracts, opaque pricing, no hedging, no transparent forward curve. Enterprise buyers commit nine-figure annual budgets entirely unhedged.
H100 one-year rental prices rose 40% in five months, from $1.70/hr to $2.35/hr. Enterprise buyers had no institutional venue to hedge this exposure.
On-demand GPU capacity sold out across all types. Blackwell generation fully booked through late 2026. Marginal compute almost unobtainable.
Three competing indices, very few transactions, and no integrated offering for index access, a trading venue, and settlement. No single participant has assembled all the pieces.
Every major commodity class developed derivatives within a decade of hitting $100B in annual trade. GPU compute crossed that threshold in 2024.
The same structure as crude oil in 1983: producers who want to lock in revenue, consumers who want to lock in costs, and speculative liquidity drawn by volatility.
TAM: $200B+ GPU compute market growing at triple-digit rates. Derivatives markets typically reach 10–20x the underlying physical market value.
Cash-settled forwards and swaps referencing published GPU benchmark indices. Voice-brokered initially — the same model that launched crude oil, natural gas, and power derivatives. Settlement fees paid in a native utility token on Canton.
Buyer locks in a fixed GPU-hour rate; seller pays floating index rate. The buyer's effective cost is fixed. The standard entry point in every commodity OTC market — energy, metals, agriculture all started here.
Agreement to buy or sell GPU capacity at a fixed price on a future date. Cash-settled against the published benchmark index at expiry. One-year tenor as starting point — the segment showing fastest price sensitivity.
Settlement on the arithmetic average of daily index values over the contract period — matching how compute is consumed as a continuous flow rather than at a single point. The most natural fit for the commodity.
Public blockchains expose positions. Centralised exchanges lack programmability. Canton provides trade-level privacy and native financial contract modelling — the two requirements that no other infrastructure satisfies simultaneously.
Positions, counterparties and trade sizes never exposed. No institutional participant will trade derivatives on a platform with public visibility. A regulatory and competitive prerequisite.
Native financial contract lifecycle: creation, novation, exercise, settlement, margining — with the precision that ISDA documentation achieves in legal prose. Purpose-built for financial instruments.
All legs of a transaction succeed or fail together, eliminating settlement risk. Essential as the market matures toward complex multi-leg structured products — options, capacity swaps, spread instruments.
Goldman Sachs, BNP Paribas, Deloitte, Microsoft already on the network. Their compliance teams have pre-approved the infrastructure — a new Canton-based product activates existing institutional machinery.
From spot market to index to OTC forwards to exchange futures to options. This is how crude oil, natural gas, power, and interest rates all developed. GPU compute follows the same path.
Partner with existing benchmarks — Silicon Data SDH100RT (Bloomberg, DRW/Jump-backed), SemiAnalysis GPU Pricing Index (9 GPU types, contract-market focus), Ornn OCPI (transaction data).
Voice-brokered forwards & swaps between known counterparties. Canton settlement with utility token for transaction fees. Cash-settled against published indices. This is what we are building.
Standardised, cash-settled monthly contracts once the OTC market establishes a reliable forward curve. CFTC/FCA regulated. Central clearing via established CCPs (LCH, CME, ICE).
Capacity swaps, range accruals, GPU price floors for AI lab budget planning, compute-vs-power spread instruments. Investment banks engage as dealers. Revenue multiples increase.
Revenue model: transaction fees on each settlement, paid in native utility token. The OTC venue captures economics at every phase of market development.
The market requires four elements: a benchmark index, OTC intermediation, settlement infrastructure, and capital markets distribution. No single participant has all four. Capacity Derivatives on Canton fills the gap.
| Player | Index | OTC Trading | Canton | Distribution |
|---|---|---|---|---|
Ornn $5.7M seed — indices + first swap, pursuing CFTC DCM licence | ||||
OneChronos / Auctionomics $32M raised — combinatorial auction, proprietary opaque standardisation unit | ||||
Silicon Data $4.7M from DRW + Jump — SDH100RT index on Bloomberg terminals | ||||
SemiAnalysis GPU Pricing Index + API — 9 GPU types, 180K subscribers, contract-market focus | ||||
Capacity Derivatives OTC brokerage on Canton — derivatives trading + quant + technical build |