Arthur Hayes Proposes FLOP Spot Rate and FRCR Compute Futures
Flop would calculate a spot-compute rate that Hayes says could become a global benchmark if the network serves AI agents. Proposed derivatives would use an hourly, volume-weighted FLOP/GigaFLOPS index.

BitMEX co-founder Arthur Hayes proposed a system for pricing AI computing capacity that would calculate a spot rate and use a separate index for futures and perpetual swaps.
Hayes says Flop Network’s FLOP-to-compute rate could become a global benchmark for spot computing if Flop becomes a payment network for AI agents. He argues that computing capacity lacks a unified global price and that chip and AI token prices derive from the cost of compute.
The network would produce a block every second. Receipts in each block would record inference latency, floating-point operations and transaction fees, which would be used to calculate the spot rate.
The proposed derivatives would use the Flop Reference Compute Rate (FRCR), an hourly, volume-weighted $FLOP/GigaFLOPS rate. Hayes outlined a cash-settled FRCR Quanto Quarterly Futures Contract and an FRCR Quanto Perpetual Swap Contract, with U.S. dollars used as margin.
Users could reserve computing capacity for delivery up to 30 days ahead. A qualifying reservation would create a tradable Flop Compute Unit (FCU), similar to a deliverable forward. Market makers could build long or short exposure by buying or supplying FCUs and rolling them forward.
Hayes previously said the FLOP testnet was expected to launch within weeks. TokenPost covered the project’s testnet plan.