ZAI allocation and supply policy

Every fee lands in the treasury in stablecoins. The other asset, ZAI, is allocated, governed and tied to use of the network on fixed on-chain mechanics: 100M genesis supply, 1B hard cap, DAO-only minting through Governor + Timelock, and a token whose job is collateral + governance while all payments stay in stablecoins (tokenomics.md, economics.md).

Principles (what the token is and is not)

Four principles. ZAI is a security budget, not a currency: its market value collateralizes the marketplace, in operator stakes, juror stakes and escrow caps, and the design targets durable demand from utility, not trading velocity. Real yield only: delegators and operators earn stablecoin revenue share, the healthiest yield in crypto; token emissions are a bootstrap tool with sunsets, never the product. No low-float/high-FDV theater: value is communicated against the 100M genesis, and the 1B cap is a decade of optional headroom, not a target. Conservative language, plain mechanics: no profit-promising language, no holder "dividends", and the buy-&-burn programme is framed and executed as protocol supply policy.

Genesis allocation (100M ZAI)

The genesis supply is split into five buckets, each with its own vesting and control:

The ecosystem bucket (40M) breaks down into 15M protocol-owned delegation, the flagship mechanism; 10M usage airdrops, points-based and sybil-filtered; 10M grants/integrations for service templates, SDK ecosystems and framework plugins; and 5M juror incentives to bootstrap arbitration. Circulating supply at genesis is ≈ 10%: the 5M liquidity plus roughly 5M of the first airdrop tranche; everything else vests or sits behind DAO votes.

The flagship mechanism: protocol-owned delegation (POD)

Node entry is $5–10k of stake for sybil resistance, yet the network wants a long tail of independent operators, because an agent that tries one service against another needs somewhere to switch to. POD resolves that tension: instead of cash grants, the treasury delegates ZAI to vetted new operators through the delegation machinery the protocol already has. The operator gets capacity without capital, because escrow caps scale with total stake. The treasury earns the delegator share of their real USDC revenue, so the incentive is recoverable, even profitable, unlike a grant. Slashing risk does the vetting: treasury delegation to a cheater gets diluted, so the program self-polices and the DAO curates. As an operator earns, they self-stake and the POD rotates to the next cohort, so 15M ZAI recycles indefinitely.

Emissions policy (the 900M headroom)

Nothing is scheduled. Minting is an explicit DAO vote through the timelock, per program, with a KPI and a sunset. The published guardrail is ≤5% of current supply per year, and never to backfill yield that real fees should provide. If most of the 900M is never minted, good; the cap is flexibility for a decade of unknowns, such as migration incentives or massive-scale juror pools, not a promise of dilution.

Value accrual (how demand meets supply)

Demand is utility-driven and enforced by the contracts: every operator locks ~$7.5k of entry stake, and more to scale escrow; every juror locks stake to earn arbitration income; every delegator locks ZAI to earn USDC real yield; and governance sets parameters people care about. Supply is throttled by vesting and by staking locks (7-day unbonding).

The treasury flywheel closes the loop. The 5% protocol fee on every settlement, the 1% conversion fee on every deposit that arrives from another chain or asset, 80% of slashes and the managed-tier fiat fees accumulate in stablecoins, and the DAO spends them on audits, juror reward pots, POD and the burn programme.

Buy-&-burn. A governance-tunable share of protocol fee revenue, 30% by default, range 0–100%, market-buys ZAI each epoch and calls burn() (ERC20Burnable; burns permanently reduce supply under the immovable 1B cap, covered by test). Usage creates buy pressure, buy pressure creates deflation, all funded by real fees, never by emissions. The slash-burn kicker: 25% of the treasury's slash proceeds are burned, so fraud literally shrinks the supply. No contract changes: it is treasury policy executed through governance.

What drives the value — in one breath

ZAI is capacity. ~$7.5k of stake puts a node on the network, but that only buys a baseline: every additional concurrent customer channel requires more staked ZAI (tokensPerSession), and the total client money a node may hold is capped at ~50% of its stake's value (the escrow cap). A node serving 10× the customers needs roughly 10× the collateral, which is exactly what protects those customers: more concurrent connections mean more stake slashable if the operator cheats. Growth and safety are the same number. As the marketplace scales, demand for ZAI scales with it, not by narrative, by require().

Operators who can't afford their own capacity don't hit a wall: anyone can stake ON an operator (delegation) and earn a share of that operator's real USDC fees, accepting a slice of the slash risk. Capacity gets crowdfunded by people betting on good operators, and bad operators can't attract stake, not least because the agents that use them rate them on-chain after every settled session. Protocol-owned delegation is the treasury doing the same thing.

On the supply side, a share of protocol fee revenue market-buys ZAI and burns it: deflation funded by real usage, under a hard cap that never moves.

The mechanics in detail

ZAI is a work token with an enforced collateral ratio: the contracts create demand that scales with marketplace throughput, and every core use locks supply rather than spending it.

1. The escrow-cap bind (the deepest driver). PaymentManager refuses new sessions once an operator's open client escrow exceeds ≈ k (≈ 50%) of their stake's value. Summed over the network, that is a protocol-enforced inequality:

total concurrent escrow  ≤  k × (market value of staked ZAI)
        ⇒  staked ZAI value  ≥  concurrent escrow ÷ k

Illustrative arithmetic: $10M of concurrent client escrow at k = 0.5 requires ≥ $20M of ZAI staked, or sessions start bouncing and operators must buy or attract more stake. Marketplace growth mechanically pulls collateral demand; the token's floor story is tied to GMV, not sentiment.

2. The real-yield bid. Delegators lock ZAI to earn the delegator share of gross stablecoin revenue (~10% by default). That gives ZAI a discountable-cash-flow anchor for lockers: if the network settles $100M/yr, ~$10M/yr flows to delegators, and capital bids ZAI (and locks it) until that yield normalizes against comparable risk. More fees mean either more ZAI locked or a higher price, and both tighten the security budget.

3. Structural buy-flows. Operator entry stakes ($7.5k each), juror stakes ($2–5k each), and the treasury's procurement buybacks, in which USDC fee revenue market-buys ZAI to fund juror reward pots, together make a recurring buy programme sized by protocol income.

4. Supply throttles. Every core use is a lock (staking, delegation, juror bonds; 7-day unbonding), vesting gates insiders, there is no emissions faucet creating perpetual sell pressure, slashing moves tokens from bad actors to the treasury, and POD keeps the ecosystem allocation working, delegated and earning, instead of liquid. Velocity, the classic utility-token killer, is structurally low because ZAI is held to hold, not to spend; the spending currency is USDC by design.

The flywheel: more agents paying leads to more concurrent escrow, the cap binds, more ZAI is staked or bought, the security budget grows, the marketplace gets safer, and more agents pay. In parallel, more fees mean richer delegator yield and bigger treasury buybacks, and so more ZAI locked.

Each failure mode has its mitigation. Reflexivity down: a falling price shrinks stake value, so escrow caps tighten, safe but capacity-crunching, or, if governance ignores it, the network is under-collateralized; hence governance retunes the ratio on drawdowns and k carries margin. Usage is the driver: if marketplace usage falls, none of this binds, and the token tracks the marketplace's use by construction. Loose parameters: if governance sets k or minimum stakes too generously, collateral demand decouples from throughput; the guardrail is that stakers themselves are the voters with the most to lose from a weak security budget.

Security parameters

Governance sets the collateral parameters against the ZAI price, on the reasoning laid out in economics.md: minNodeStake at about $7,500 of ZAI of total stake behind a node (the operator's own floor, minSelfStake, a small fraction of it), escrow caps at k ≈ 0.5 for every accepted asset, a slash floor of about $500 beneath the 10% proportional slash, a juror minimum of about $2–5k, and a quorum of at least 3–5× the largest expected single juror.

Those numbers are the security budget the network runs on.

Next: Operating the network — what runs on-chain and off when the network is live, and how it is watched and governed.