Start with a problem you already know.
ChatGPT. Gemini. Claude. These are products built by large corporations that spent hundreds of millions of dollars training the AI models inside OpenAI, Google, Anthropic. But that price tag also means that only a handful of companies in the world can afford to build cutting-edge AI. Those companies decide what the AI can and can’t do. They control access to it, set the prices. They own it entirely. And they capture almost all the value it creates.
Bittensor’s founding question was simple: what if you didn’t have to do it that way?
What if, instead, you could build AI the way Bitcoin built money? Distribute it across thousands of independent participants. None of them owns the whole thing. All of them are rewarded for contributing to it.
That is, in one paragraph, what Bittensor is.
The Bitcoin Comparison
Most people know that Bitcoin is a decentralized currency. It’s not printed by any government and no bank controls it. Instead, thousands of computers around the world run software that collectively agrees on who owns what. The people running those computers, the miners, are rewarded in Bitcoin for doing the work of keeping the network honest.
Bittensor borrows that structure but points it at something different. Instead of reaching agreement on financial transactions, the network reaches agreement on the quality of intelligence. Instead of rewarding miners for processing transactions, it rewards them for producing useful AI outputs.
The native currency, the token that flows through the whole system, is called TAO. Like Bitcoin, TAO has a fixed maximum supply: 21 million coins, ever. No central bank, no government, no company can print more.
The Basic Unit: A Subnet
The core building block of Bittensor is the subnet. Think of a subnet as a specialized marketplace for a specific kind of digital work.
Take a simple example. Imagine a subnet whose job is: answer medical questions accurately. Inside that subnet, there are two kinds of participants:
The miners who do the work. Each miner runs an AI model. When a question comes in, every miner produces an answer.
Then, the validators judge the work. They look at all the answers from all the miners and score them: which was most accurate, most useful, most complete?
Those scores go back to the blockchain, which uses them to decide how much each miner earns. Better answers, more income. If a miner contributes consistently poor answers, he eventually gets pushed out as new competitors register.
Marcus Graichen, founder of Taostats, the network’s leading analytics platform, describes it this way: “When you ask ChatGPT a question, it’s like having one university professor in front of you. What Bittensor does is fill a room with a hundred professors, ask the question to all of them, grade those answers, and give you back the best.”
Why Have Subnets? 128 of them?
Because different problems need different kinds of work.
One subnet produces AI text responses. Another handles image generation. Another provides blockchain infrastructure services to developers. Another analyses financial markets. Another researches pharmaceutical molecules.
Right now there are 128 active subnets on Bittensor, but the network’s roadmap targets to increase it to 256 by end of 2026. Each subnet has its own economy, its own rules for what counts as good work, and its own community of miners and validators.
TAO and Alpha: The Two-Token Economy
This is where the economics get more layered, but it is worth following.
Originally, the network had one token: TAO. All rewards across all subnets were paid in TAO. In April 2025, Bittensor introduced a system called dTAO (dynamic TAO) which changed the system fundamentally.
Under dTAO, every subnet gets its own token, called an alpha token. Each alpha token has again a fixed maximum supply of 21 million, exactly like Bitcoin, and its own emission and halving schedule, independent of TAO’s.
Here is how the two relate:
Alpha tokens are what miners and validators within a subnet earn for their work. The price of a subnet’s alpha token reflects demand for that subnet’s services. If lots of people are using it and investing in it, the alpha token rises in value.
TAO is the reserve currency of the whole economy. It flows into each subnet’s liquidity pool based on how much demand the subnet attracts. More demand, more TAO flowing in, higher value for the alpha token.
Think of it this way: TAO is the dollar. Alpha tokens are the local currencies of individual businesses operating within the same economy. Connected, but able to move independently.
Staking: How You Participate Without Mining
You do not have to run a miner or a validator to participate in Bittensor. You can also be a delegator who holds TAO and assigns it to a validator you trust.
When you delegate your TAO to a validator, that validator uses your stake as part of their weight in the network. In return you earn a share of what that validator earns. Validators typically charge a small percentage of earnings, their “take”, to cover their costs and team.
This delegation model is what funds Taostats. People who use the platform every day choose to delegate their TAO to the Taostats validator. That delegation generates the revenue Mog uses to pay his team and run the infrastructure — no venture capital, no advertising.
The Halving
Both TAO and each subnet’s alpha token follow a halving schedule. At regular intervals, the number of new tokens produced per day is cut in half, the same mechanism Bitcoin uses to manage supply over time.
Bittensor’s first TAO halving happened in December 2025. Daily emissions dropped from 7,200 TAO to 3,600. Because miners are now paid in alpha tokens instead of TAO directly, the TAO halving had less immediate impact on day-to-day miner economics than many expected.
The more consequential event for each subnet will be its own alpha halving, when that subnet’s token emissions get cut in half. The subnets that are already generating real revenue from real customers by that point will be fine. Those still depending entirely on token emissions will face a harder test.
Why Institutions Are Paying Attention
The network generated over $43 million in revenue across its subnets in the first quarter of 2026 alone. Grayscale filed with the SEC to launch a spot ETF around TAO. Asset managers who were ignoring this space two years ago now have analysts covering it.
The reason is not hype. It is that Bittensor represents a testable thesis: that distributing the ownership and economics of AI infrastructure across thousands of independent participants produces better, cheaper outputs than a single centralized provider. And the people doing the work capture the value, rather than having it flow upward to a small number of shareholders.
Whether that thesis holds at scale is still being established. But the early numbers suggest it is worth taking seriously.
