​

Abstract

We present the tokenomics framework for $AVM, an on-chain compute protocol enabling decentralized execution nodes to provide AI inference and code execution as a service. The design aligns incentives through a phased distribution schedule and a self-sustaining economic model. In Phase 1, a transaction tax funds node deployment; in Phase 2, diminishing token rewards encourage decentralization; and in Phase 3, all compute payments flow directly to node operators, who may run and monetize their own hardware.

​

1. Token Distribution

The total supply of $AVM is defined as T_total = 100,000,000 (one hundred million)

Let T_total denote the total supply of $AVM tokens. Allocations are defined as follows:

​

2. Phased Economic Model

​

Phase 1: Bootstrapping (0 → 6 months)

• Transaction Tax: A 5% fee on each $AVM trade.
• Allocation of Fees: All proceeds are directed to:
  1. Deploy and initialize execution nodes.
  2. Build liquidity reserves for market stability.
  3. Fund early developer and infrastructure grants.

This phase ensures that network usage immediately translates into expanded compute capacity, obviating reliance on external funding.

​

Phase 2: Reward Distribution (6 → 12 months)

• Node Operator Incentives
  • Active execution nodes receive $AVM rewards on a linearly decreasing schedule, defined by:

    R(t) = R0 × (1 − t / 6 months)   for  0 ≤ t ≤ 6 months
    

  • This schedule encourages early deployment and geographic/provider diversity.
• Transition Mechanism
  • As the reward R(t) approaches zero, node operators obtain revenue exclusively from compute fees paid by AI agents.

​

Phase 3: Self-Sustaining Economy (≥ 12 months)

• Compute Payments
  • AI agents submit tasks priced in $AVM or fiat via the x402 protocol.
• Decentralized Node Monetization
  • Any user may deploy an execution node:
    • Revenue Model: Operators earn proportionally to compute utilization.
    • Payout Options: Earnings can be disbursed in $AVM or fiat.

​

3. Decentralized Node Participation

After Phase 2, the protocol fully decentralizes compute provisioning:

1. Node Deployment
   • Users provision hardware (CPU/GPU) and run the AVM client.
2. Service Registration
   • Nodes register on-chain, advertising available capacity and pricing.
3. Task Execution
   • AI agents select from registered nodes via a discovery mechanism.
4. Revenue Distribution
   • Compute fees are automatically settled on-chain according to:

     Pi = (Ui / Σ Uj) × F_total
     

     where Ui is an individual node’s usage and F_total is the total fees collected.

​

4. Conclusion

The $AVM tokenomics design integrates vesting-aligned allocations with phased incentives to bootstrap, decentralize, and sustain a global network of AI compute providers. By enabling open participation in Phase 3, AVM transforms compute from a centralized bottleneck into a distributed, user-driven economy.