Pays a global network to burn energy on math puzzles → “transaction confirmed.” Nothing else.
Pays the same kind of network to produce real machine intelligence people actually use.
Bitcoin proved you can pay millions of strangers in crypto to power one network. Bittensor keeps that incentive engine and abstracts it from “provide energy” to “provide any valuable work.”
Bitcoin pays a global network to burn energy securing one ledger. Bittensor uses that same incentive engine, but pays its network to produce useful intelligence: answering queries, generating media, forecasting, and more. Bitcoin was money. Ethereum was apps. Bittensor is intelligence.
TAO is Bittensor's base money, and it borrows Bitcoin's tokenomics almost exactly: a fixed 21 million supply, no pre-mine, and a halving schedule (the first halving hit in December 2025). The difference is how you earn it: not by burning energy, but by producing valuable AI work. This is the asset our fund holds as its strategic core.
Each subnet is effectively a startup living inside Bittensor, specialized in one job: image generation, price prediction, GPU compute, scientific forecasting, and far more. There are 128 live today, each producing its own "digital commodity." This is where the venture-style upside lives, and where our diligence is focused.
Every subnet has its own token, its alpha token, and the open market decides which subnets win. Capital stakes into the subnets it believes in; more stake means more daily TAO rewards flowing to that subnet. The strongest subnets now go further: they generate real revenue from their product and use it to buy back their own token, linking token value directly to usage. That mechanism is central to how we pick what to own.
Inside each subnet, miners do the actual AI work and validators grade its quality. Rewards follow the grades, scored by an algorithm called Yuma Consensus, a self-running, always-on quality contest. Good work is paid; bad work is starved. No central operator decides who wins.
Bittensor applies a proven crypto model, paying a decentralized network to produce valuable work, to artificial intelligence, a market today concentrated in a small number of companies. We believe disciplined exposure to the network, taken early and sized appropriately, can compound over a multi-year horizon.