AI Agents Need Their Own Money to Pay for Computing Power

Arthur Hayes, co-founder of BitMEX and founder of Flop Labs, has proposed an innovative solution for AI agents to pay for computing power directly using a token. In his recent project, Flop Labs, Hayes argues that AI agents need a unique form of money that can be exchanged directly for computing resources, rather than relying on traditional payment methods.

The problem with current systems is that they are designed for human users, not autonomous agents. Agents don't have the same needs as people; they consume floating-point operations (FLOPs), which are the calculations required to run AI models.

Hayes' idea is to create a currency that can convert into compute on demand at a fair price. This would be particularly useful for AI agents, who require a constant supply of computing power. The proposed Flop Network aims to provide this solution by allowing GPU operators to receive FLOP tokens in exchange for providing inference services.

The project's tokenomics are still preliminary, but the plan is to create 17.2 billion FLOP tokens by year 10, with a Genesis airdrop of 3.5 billion tokens allocated among miners, agents, validators, and reserves.