Bittensor (TAO) Explained: Subnets, Yuma Consensus and a Market for AI
Not affiliated with this project. We received nothing for this page. Not financial advice.
Quick answer
Bittensor is a network that pays people in TAO tokens for providing useful AI and machine-learning work. The network is split into subnets, each focused on one task (text, images, prediction, compute and more). Miners do the work, validators score it, and Yuma Consensus turns those scores into rewards. TAO has a 21 million supply cap and halvings, like Bitcoin.
| Token | TAO, max supply 21 million |
|---|---|
| Structure | Subnets, each with its own task and incentives |
| Roles | Miners (do work), validators (score work) |
| Consensus | Yuma Consensus for reward distribution |
| Upgrade | Dynamic TAO: each subnet has its own token |
| Emissions | Halvings, the first in late 2025 |
The big idea
Big AI labs train models behind closed doors. Bittensor tries to create an open market where anyone can contribute models or computing power and get paid according to how useful their contribution is, as judged by the network.
Subnets, miners and validators
Each subnet is like a mini-marketplace for one job. Subnet owners define the task and how it's measured. Miners submit outputs, and validators test them and assign scores. Rewards flow to the miners and validators whose work the network rates most highly.
Yuma Consensus
Yuma Consensus combines validators' scores, weighting validators by stake and penalising those whose scores disagree with the consensus. This aims to stop a single validator from rewarding its own miners unfairly.
Dynamic TAO
With the dynamic TAO upgrade, each subnet got its own token (alpha). Staking TAO into a subnet buys its alpha, and the market's demand for each subnet helps decide how emissions are split. This made subnet selection more market-driven, but also added a layer of speculation.
TAO supply
Like Bitcoin, TAO has a 21 million cap and emissions halve over time. The first halving cut daily issuance in late 2025. Fewer new tokens can support price only if demand for the network holds up.
Main risks to know
- Hard to verify whether subnets produce AI that customers actually pay for.
- Validator and stake concentration.
- Subnet tokens add complexity and speculation.
- Strong competition from centralised AI providers.
Frequently asked questions
What is a Bittensor subnet?
A specialised network inside Bittensor with its own task, miners, validators and incentives.
How do you earn TAO?
By mining (providing useful work in a subnet), validating, or staking TAO with validators.
What is the TAO supply cap?
21 million, with halvings that reduce emissions over time.
What is dynamic TAO?
An upgrade that gave each subnet its own token and let market demand influence emission splits.
Sources
- Bittensor documentation · Subnets, Yuma Consensus, dTAO
- Bittensor · Project site and whitepaper
- TAO on CoinMarketCap · Market data
Figures such as prices, supply and user numbers change quickly. We cite the source and date; check the live figure before you act.