A network where subnets produce AI commodities and the chain pays for verified work.
Live market data
CoinGecko data; may be delayed.
Tokens remain locked or unissued: today's price discounts a larger future supply.
45.7% of the achievable supply is already circulating.
$757.6 -68.71% · Mar 7, 2024
Source: CoinGecko, snapshot of Sep 13, 2026. Figures may be delayed.
This asset is not mapped to a DefiLlama-tracked protocol or chain, or the provider reports no data for it. When a verified mapping exists, the table appears here.
Bittensor is a proof-of-stake network whose product is a market for machine intelligence. Independent subnets — machine-learning pipelines competing on quality — produce commodities like inference, training compute, data, and prediction, and validators rank their outputs on-chain. TAO is emitted to reward the contributors whose work the network actually scores as valuable, which makes the token a claim on other people's compute rather than on a company.
TAO is a claim on a market for AI work, not on a company or a chain's gas.
The Bitcoin-like cap gives the supply story discipline most AI tokens lack.
The thesis lives or dies on whether subnet output has buyers outside the network.
Subnets are the core unit: each runs a specific machine-learning task with miners producing outputs and validators scoring them. Scoring follows Yuma consensus, which converts validator agreement about output quality into emissions; subnets that produce garbage earn nothing, which is the mechanism that makes the network a market rather than a grant program. Root-network validators allocate TAO emissions across subnets, so capital flows toward the subnets whose commodities the network prices highest.
Staking to validators who score subnet quality
Registration and incentive budget inside subnets
Unit of payment for machine-learning commodities
TAO's value capture is unusual: emissions are the supply schedule and the incentive budget at once, so every token that exists was paid for work the network scored. Whether that translates into durable demand depends on whether the commodities subnets produce — inference, data, prediction — are worth buying off-chain, not just scoring on-chain. That is the honest open question: on-chain scoring of AI output is verifiable; off-chain willingness to pay is not guaranteed to follow.
TAO's schedule mirrors Bitcoin's: a 21 million hard cap with halvings, a fixed emission curve, and no premine allocated to a corporate treasury. Emissions currently flow to miners, validators, and subnet owners according to on-chain scores. The cap plus the burn of unclaimed rewards means supply discipline is enforced by code, not by policy.
Protocol changes go through OpenTensor Foundation's improvement proposals, and the foundation funds core development. Subnet registration is permissionless, but the root validators' allocation of emissions is a de facto governance lever: they decide which subnets the market funds. Critics note this concentrates influence in a small validator set; supporters answer that it is the same concentration any market's pricing committee has.
Demand for subnet output is unproven outside the network's own emissions loop
Validator concentration in root-network allocation is a de facto control point
Scoring mechanisms can be gamed; Yuma consensus is patched in response, but exploits have happened
Emissions create continuous sell pressure from recipients who earn in TAO and pay costs in fiat
Not directly, and the comparison is usually misleading. Centralized labs own their compute and sell inference; Bittensor coordinates independent, unowned compute and pays contributors per unit of scored output. It competes for the same underlying resources — GPUs, data, talent — but the product is a market layer, not a closed model.
That on-chain scoring substitutes for real demand. If the subnets' outputs are only valuable inside the network's own scoring game, emissions become a closed loop: tokens are paid for work only the network itself buys. Watching whether external applications actually consume subnet output — and pay for it outside emissions — is the signal that separates a market from a subsidy.
No verification date recorded.
No citations recorded for this profile yet. The official links above are the starting point.
Pending linkage to a concrete organization on the map.
Serves a comparable role, so the two are worth reading side by side.
Directory suggestion, pending verification.
Serves a comparable role, so the two are worth reading side by side.
Directory suggestion, pending verification.
Adjacent asset in the same research context.
Directory suggestion, pending verification.
