Hayes outlines FLOP AI compute derivatives plan after altcoin exit

Arthur Hayes, BitMEX co-founder and CEO of Flop Labs, has laid out a more concrete framework for how the Flop Network’s native token could help turn AI compute into something closer to a tradable commodity. Writing two days after selling portions of his altcoin holdings, Hayes argued that if AI agents use Flop as a payments layer, then a measurable “compute spot price” denominated in the network token could become the market’s global reference point. In a Monday Substack essay titled “The Trifecta,” Hayes proposed valuing AI work by tracking the computing power delivered per inference task and then using on-chain transaction data to compute a spot rate for compute priced in Flop (FLOP) tokens—an approach he said is needed because there is currently no universally accepted benchmark for AI computing power. Key takeaways Hayes says Flop’s payments role for AI agents could make a FLOP/compute exchange rate a widely used benchmark for spot compute. The proposal relies on measuring compute delivered per inference and deriving pricing from transaction-level data. Hayes links the development of a liquid spot compute market to the possibility of derivatives built on that reference price. Flop Labs has discussed plans for a large token distribution in Q4 2026, alongside its broader push for compute payments and decentralized memory services. The timing of the explainer follows reports and on-chain tracking highlighting sizable sales by Hayes-linked wallets. From “AI compute” to a spot price Hayes’ core claim is that markets need a consistent unit of measurement to price AI compute in the same way that exchanges price conventional commodities. In his framing, Flop becomes the rails through which AI agents pay for compute, and the key market output becomes the FLOP/compute exchange rate. “If Flop becomes the payments network for AI agents, then it follows that the $FLOP/compute exchange rate will be the global benchmark for spot compute,” Hayes wrote in the essay. He added that a liquid spot compute market would allow derivatives to reference that benchmark, effectively extending the same pricing-and-risk machinery that exists in traditional commodity markets. Hayes also emphasized a practical gap in today’s ecosystem: without a broadly accepted “benchmark price” for AI computing power, it’s difficult to standardize what compute is worth across providers, workloads, or networks. Why a compute commodity framework matters The push to make compute tradable sits within a larger debate about how AI infrastructure should be monetized. In a July 2026 report, BCG examined how quickly AI computing power could shift from being priced in fragmented, provider-specific ways to becoming more commodity-like. According to BCG’s projections, the AI compute market could grow to roughly $2.3 trillion by 2030, rising from about $360 billion in 2025.