What happened
Moca Network published the new mechanism on Thursday, according to CryptoBriefing's report. Stakers holding Staking Power, the non-transferable balance accrued by locking the project's native asset, can now burn that balance to mint LLM tokens. Those LLM tokens are the unit of payment for inference calls made by AI agents running against models integrated with Moca's identity stack.
The flow runs one way. Staking Power goes in, LLM tokens come out, and the LLM tokens are consumed when an agent makes a call. There is no path back to staked principal once burned. CryptoBriefing framed the move as an attempt to redefine digital identity's role in AI economics, tying token utility to resource allocation and consumption.
The initial coverage did not include conversion ratios, a cap on weekly mints, or fee splits between model providers and the protocol treasury. Those details will decide whether the burn meaningfully tightens supply or remains symbolic.
Why it matters
Most staking economies pay holders to do nothing. Lock the token, earn the yield, exit when the unlock clears. Moca's design inverts that. Staking accrues a balance you spend, not a balance you compound. The exit is a burn, and the burn pays for compute.
That matters for two reasons. First, it ties token demand to actual usage of the AI agents the network is positioning itself around. If the agents see traction, the burn pulls supply. If they do not, the staking layer offers nothing else to hold for. Yield-free staking is a high-conviction design choice.
Second, it makes identity a billing primitive. Moca's stack was built around persistent crypto-native identity. Connecting that identity to a metered, burnable resource means every agent action carries an authenticated user behind it. Regulators chasing AI accountability frameworks will notice.