2025-05-20·9 min read

The Economics of Feedable Network

Economic design is where most blockchain projects make their worst mistakes. They either inflate the supply to fund operations, design tokenomics that primarily benefit insiders, or create fee models that misalign miners and users. Feedable Network's economic design starts from first principles: the token should reward real work, burn with real usage, and have a fixed, predictable supply.

Supply: why 21 million

The total supply of F is 21,000,000 — a hard cap with no exceptions. The choice mirrors Bitcoin deliberately. The psychology of a hard cap matters. Participants know the maximum dilution they will ever face. There is no governance vote that can change it. There is no "emergency inflation" mechanism.

At a 10-second block time with 50 F genesis rewards, the initial emission rate is 5 F per second. The halving schedule reduces this every 210,000 blocks — approximately every 24.3 days at current block times. The full supply takes decades to emit, front-loaded toward early network participants who take the most risk.

The halving schedule

Era 1: blocks 10,001 to 220,000   — 50 F per block
Era 2: blocks 220,001 to 430,000  — 25 F per block
Era 3: blocks 430,001 to 640,000  — 12.5 F per block
Era 4: blocks 640,001 to 850,000  — 6.25 F per block

The genesis phase (blocks 0 to 10,000) is a special bootstrap period using Proof of Stake with 30 F per block. This gives the network time to establish its validator set before Proof of Feed becomes mandatory at block 10,001.

The halving creates predictable supply reduction. Miners who operate early earn more tokens per unit of work. This compensates for the higher operational risk of running early-stage infrastructure.

Genesis allocation

The 21 million F is split at genesis:

60% (12,600,000 F) to mining rewards — released via the halving schedule over years
15% (3,150,000 F) to team and founder — 3-year lock, monthly vesting
10% (2,100,000 F) to genesis validators — 2-year lock, monthly vesting
10% (2,100,000 F) to treasury — governed by the validator set
5% (1,050,000 F) to investors and advisors — 1-year lock, monthly vesting

The vesting locks are enforced on-chain via TigerBeetle scheduled releases. There is no manual disbursement. The team allocation cannot be accelerated by any governance decision during the lock period.

The 60% mining share is the largest allocation by design. The network exists to serve AI inference. Miners who do that work should capture the largest share of the token supply.

Fee model: EIP-1559 with a burn

Feedable Network uses an EIP-1559 style fee model. Every transaction has a base fee that is auto-adjusted per block based on demand. When blocks are full, the base fee rises. When they are empty, it falls. This creates a predictable fee market without the bidding wars of first-price auctions.

The base fee split is 50% burned permanently and 50% paid to the block miner as a tip. The burn is deflationary. As network usage increases, more F is destroyed. If usage is high enough, the burn rate can exceed the emission rate, making F net-deflationary.

For inference jobs specifically, the fee formula is:

fee = data_size_kb × priority_multiplier × base_fee

The priority multiplier reflects the SLA tier. Critical jobs (sub-millisecond requirement) pay 10x the base fee. High priority (sub-5ms) pays 3x. Normal (sub-50ms) pays 1x. This creates a market for latency — users who need faster inference pay more, miners who can deliver faster inference capture that premium.

Block reward split

When a block is finalized, the 50 F reward (in Era 1) is split:

70% to the miner who produced the valid FeedProof
20% to the validators who confirmed the block via BFT voting
10% to the treasury

The miner receives the majority because they do the most work — running Blazil, generating the proof, proposing the block. Validators receive a smaller share for confirming. The treasury accumulates F that the validator set can deploy for protocol development, security audits, or ecosystem grants.

Why the numbers work

The economic model has three properties that make it sustainable.

First, the burn mechanism aligns token value with network usage. More inference jobs means more base fees means more F burned. The supply decreases as the network becomes more useful, not in spite of it.

Second, the SLA enforcement creates accountability. A miner who accepts a Critical SLA job and delivers Normal latency faces automatic fee refunds and stake slashes. There is no way to accept premium fees and deliver standard service. The protocol enforces the contract.

Third, the vesting locks prevent early participants from dumping. The team, genesis validators, and investors all have multi-year locks with monthly vesting. Their incentives are aligned with the long-term success of the network, not the short-term token price.

The mining rewards — 60% of total supply, released over decades — ensure that the network can always attract miners with meaningful incentives. Even when the block reward halves to fractions of F, the fee income from a high-throughput inference network should be sufficient to sustain operations.

Whether that plays out depends on adoption. The economic model creates the right incentives. The product has to earn the adoption.