BitMEX co-founder Arthur Hayes has proposed cash-settled contracts tied to an hourly measure of the FLOP token against AI computing capacity, a design that would connect crypto derivatives with the cost of running AI workloads.
His proposal centers on the Flop Reference Compute Rate (FRCR), an hourly benchmark weighted by trading volume and expressed as $FLOP/GigaFLOPS. That figure would anchor a quarterly futures contract and a perpetual swap, both described as quanto products. The contracts would settle in cash, with U.S. dollars used for margin. The FRCR is distinct from the network’s proposed spot rate: one would serve as the derivatives benchmark, while the other would express FLOP relative to compute.
Under the proposal, Flop would add a block every second. Task receipts would capture inference time, floating-point operations and fees, which the network would use to calculate its FLOP-to-compute spot rate.
Hayes sees a possible reference price emerging if AI agents use Flop for payments. He says computing has no common worldwide quote, while chip and AI-token prices are tied to compute expense. Hayes presented the network-based rate as a proposal.
Reservations for delivery within 30 days could produce a tradable Flop Compute Unit (FCU), which Hayes likens to a deliverable forward. Market makers could take long positions by buying FCUs or short positions by supplying them, then roll those positions forward.
The proposal comes amid a separate U.S. policy effort on AI compute futures. The Hyperliquid Policy Center urged the Commodity Futures Trading Commission to permit such contracts; the agency’s consultation did not authorize the proposed markets, as EconoTimes previously reported.
