TensorCash AI Compute Derivatives: Native, Cash-Settled Exposure to the Compute Cycle

Abstract

TensorCash ties block production to verified language-model inference: the network's mining difficulty is, in effect, a live on-chain measure of how much AI compute the world is committing to the chain. This paper specifies AI Compute Derivatives — native, cash-settled contracts whose payoff is a deterministic function of that difficulty at a committed future block. They let a miner or AI provider hedge revenue when compute floods in, and let an external party take financial exposure to the AI-compute cycle without operating a single GPU. The instrument is a margined, capped contract-for-difference: each side posts initial margin into its own Taproot vault and can lose at most that margin. A symmetric, zero-cost-to-enter form gives delta-one (linear) exposure; an optional day-one premium turns it into a covered call or put on compute. Settlement is enforced entirely by consensus — two narrow opcodes, active from genesis — with no general-purpose virtual machine, no oracle, and no intermediary. Cooperative settlement is private: on chain it is indistinguishable from an ordinary payment.