How Subnets Compete
No subnet receives a fixed TAO allocation. Each one earns a share of every block’s emission in proportion to the TAO staked into its alpha pool, and pays that share out to its own miners and validators. Attracting stake is therefore the whole game: more stake means more emission, which means miner slots are worth more, which attracts better miners, which is what justifies the stake. Run in reverse, it is equally fast — a subnet that stops delivering loses share to one that does.
How this works on-chain: the subnet modelRegistration, netuids, hyperparameters, the owner cut, deregistration and the full lifecycle — specified against the current runtime.subtensor.com/learn/core/subnetsReading the Emission Column
Emission is the most-read and most-misread column in the subnet directory. It is the percentage of the network’s newly minted TAO this subnet currently captures — not a yield, not a return, and not a promise about tomorrow. Three habits make it useful:
- Read the slope, not the level. A subnet climbing from 2% to 4% is being repriced. One that has held 8% for a year is stable, which is a different investment and a much harder subnet to mine competitively.
- Divide before you compare. Emission share is what the subnet earns; what a participant earns is that share divided by the miners, or by the staked alpha, competing for it.
- It is a market price, not a scorecard. Emission follows stake, and stake follows belief. It is the network’s current opinion of the subnet, which is worth a great deal and is not the same as evidence.
The Other Columns
Alpha price is the subnet’s token measured in TAO by its AMM pool; it moves with every stake and unstake. Market cap is that price against the alpha in circulation, so it is a valuation rather than an amount anyone could withdraw. Total stake is the pool’s depth, and therefore how much a large exit will move the price — yours or somebody else’s. Miners and validators count who is actually registered and active: many miners means a high quality bar, and many independent validators means scoring that is hard for one party to bend.
Each subnet’s own page adds the things a table cannot show: stake concentration across coldkeys, validator performance, and price history. That is where research actually happens.