Jacob Steeves
Co-founder
A former Google software engineer, known in the community as “Const”. Architect of Bittensor's incentive mechanism and its long-term vision.
About us
The story behind Bittensor
We are a training academy built around one of the boldest ideas in crypto: a decentralized market where intelligence itself is mined, measured and rewarded. To teach it well, we start where it all began.
Our mission
We train miners, validators and builders to take part in the Bittensor network with confidence — understanding not just the commands, but the economics, the risks and the history behind every block.
Origins
An idea born in 2019
Bittensor was conceived in 2019 by Jacob Steeves and Ala Shaabana, two researchers who shared one frustration: the world's most powerful AI was being built behind closed doors by a handful of corporations.
Their answer was a protocol that treats machine learning the way Bitcoin treats computation — an open network where anyone can contribute intelligence and get paid in proportion to the value they create. The original whitepaper was published under the pseudonym Yuma Rao, a nod to crypto's Satoshi tradition.
Development is stewarded by the Opentensor Foundation, while the network itself is run by thousands of independent miners and validators around the globe.
Founders
The minds behind the protocol
Ala Shaabana
Co-founder
Computer scientist with a PhD from the University of Waterloo. Led research into distributed machine learning and the protocol's scientific foundations.
Yuma Rao
Pseudonymous author
The name on the original whitepaper — and the namesake of Yuma Consensus, the algorithm that scores every miner today.
Functionality
How the network thinks
Subtensor
Bittensor's own Layer-1 blockchain, built with the Polkadot SDK (Substrate). It records stakes, weights and emissions every 12 seconds.
Subnets
Independent markets, each focused on one task — language, images, storage, data, compute or prediction. Owners design the rules; miners compete to win.
Yuma Consensus
The algorithm that combines validator scores into a single, manipulation-resistant ranking and distributes rewards accordingly.
Dynamic TAO
Since 2025, every subnet has its own alpha token. Stakers decide which subnets deserve emissions by moving TAO into them.
Networks & trajectory
From Kusanagi to Finney
2019
The concept
Jacob Steeves and Ala Shaabana start work on Bittensor; the Yuma Rao whitepaper outlines a peer-to-peer market for intelligence.
2021
Kusanagi
January: the first network goes live and TAO is fair-launched — no pre-mine, no token sale. Kusanagi is later halted to fix consensus issues.
2021
Nakamoto
November: the network restarts as Nakamoto, scaling to thousands of nodes and proving the incentive model in the wild.
2023
Finney
March: the Finney mainnet launches on the Subtensor chain, introducing the subnet architecture that defines Bittensor today.
2025
Dynamic TAO
February: dTAO gives every subnet its own alpha token and lets the market steer emissions.
2025
First halving
December: daily issuance drops from roughly 7,200 to 3,600 TAO, reinforcing TAO's scarcity.
Today
Open frontier
Over a hundred subnets compete to produce useful intelligence — and the network keeps growing.
Tokens
The TAO economy
TAO
Native asset
Hard-capped at 21,000,000. Used for staking, subnet registration and as the reward for useful work.
Alpha
Subnet tokens
Each subnet issues its own alpha token. Staking TAO into a subnet converts it to alpha through an on-chain liquidity pool.
Rao
Smallest unit
1 TAO = 1,000,000,000 rao — named in honour of Yuma Rao.
Where every block's emission goes
41%
Miners
Rewarded for the quality of their output
41%
Validators & stakers
Rewarded for accurate evaluation
18%
Subnet owners
Rewarded for designing useful markets