Bittensor Subnet Economics: How dTAO Actually Works

The old system was political

Before February 2025, a small group of root validators decided which subnets received TAO emissionsEmissionsNew tokens created and distributed by a blockchain protocol over time as rewards to validators, stakers, or miners. Emissions fund network security and participation at the cost of diluting existing holders.Like a company that pays employees partly in newly printed shares. Every year the total number of shares goes up, which means existing shareholders own a slightly smaller slice of the same company unless the company grows faster than the printing.Read more →. They set weightsParametersThe internal numbers (weights and biases) inside a neural network that get adjusted during training. A 70-billion-parameter model has 70 billion adjustable internal numbers encoding everything it has learned.Like the synapses in a human brain. Each parameter is a tiny dial that gets nudged a little during training. With enough dials, the network can represent surprisingly complex patterns. The total parameter count is roughly how much "brain" the model has.Read more → across subnets based on their judgement, and those stake-weighted votes determined each subnet’s share. It was a delegated, political process. If you knew the right validators, your subnet ate well. If you did not, good luck.

Dynamic TAO (dTAO) replaced this with markets. Each subnet now has its own tokenTokenA digital unit of value or access rights tracked on a blockchain. Tokens can represent ownership in a project, a right to use a service, a share of future revenue, or simply a tradable asset with no underlying claim.Like a physical poker chip a casino issues. The chip itself has no value. What makes it worth something is what it lets you do at the casino, what the casino has promised, and how much other people will pay you for it.Read more →, its own AMMAMMAutomated Market Maker. A type of decentralised exchange that uses liquidity pools and a pricing formula to enable token trading without an order book. Anyone can deposit tokens into the pool and earn fees from trades.Like a vending machine that sets its own prices based on how much stock is left. As one type of token gets bought and depleted, the machine raises its price for that token automatically. As the other type accumulates, its price drops. No human operator needed.Read more → pool, and its own price signal. Capital flows, not committee votes, determine who gets emissions. The system went live on mainnet around 6 February 2025.

Then in November 2025, the emission modelModelA trained neural network that takes inputs (text, images, audio) and produces outputs (more text, classifications, generated content). In DeAI the model is the thing that actually does the work.Like a very experienced apprentice who has spent years watching thousands of masters make furniture. They can't explain how they know when a joint is right, but they can make a chair that looks and functions like a Chippendale. The training is invisible. The output is what matters.Read more → changed again. The original price-based allocation was being gamed. Projects would inflate their alpha token price using TAO treasuries, collect outsized emissions during the slow price decline, and repeat. The replacement is flow-based: emissions track net capital movement into each subnet, smoothed over an 87-day window. Gaming a flow signal is harder than gaming a price.

I don’t hold TAO. Two fundamental changes in under a year. If you’re holding TAO or considering subnet participation, the mechanics matter.

Alpha tokens

Real on-chain tokens, not internal accounting. Each subnet has its own, living in a constant-product AMM pool paired with TAO.

When you stake TAO into a subnet, your TAO enters the reserve pool and the AMM calculates the equivalent alpha at the current exchange rate. When you unstake, it reverses: alpha returns to the pool, TAO exits to your walletWalletSoftware that stores the private keys needed to control tokens on a blockchain. A wallet does not actually hold any tokens. The tokens live on the chain. The wallet holds the keys that prove you own them.Like the key to a safe deposit box. The key doesn't contain your valuables. The valuables sit in the bank's vault. The key is what proves you're allowed to open the box and take them.Read more →. Minimum stake is 0.1 TAO.

SlippageSlippageThe difference between the expected price of a trade and the price you actually get when the trade executes. Slippage usually goes against the trader and gets worse with bigger trades or thinner markets.Like trying to buy 1000 bananas at the corner shop. The first few are at the marked price, but by the time you've bought them all you've moved the price up because there are no more bananas left at the original level. The shop has to restock at higher cost.Read more → is real. The docs give a worked example: a pool with 100 alpha and 10 TAO in reserves. Stake 5 TAO and you receive 33.33 alpha, not 50. That is 33% slippage. Larger stakes relative to pool size get worse. This is standard constant-product AMM behaviour, but it means entering or exiting smaller subnets carries meaningful cost.

There is also a concentrated liquidityLiquidityHow easily a token can be bought or sold without moving the price. High liquidity means you can enter or exit large positions quickly at the quoted price. Low liquidity means even small trades can swing the market.Like the difference between selling a house and selling a share of Apple stock. The house might be worth more on paper, but finding a buyer at that price takes weeks. The Apple share converts to cash in one click.Read more → layer (Uniswap V3 style) where liquidity providers can specify price ranges for capital efficiency. This adds flexibility but also complexity: you need to manage price ranges, impermanent loss, and fee collection. Not for beginners.

Each subnet’s alpha emission starts at 1 per blockBlockA batch of transactions added to a blockchain at a set interval. Each block cryptographically links to the previous one, creating an append-only chain that can't be rewritten without redoing all the work since.Like a page in a ledger. Every page has a fixed number of entries, every page references the previous page, and once a page is filled and signed off it can't be edited without visibly invalidating every page that came after. The chain is just a very long series of these sealed pages.Read more → and follows its own independent halvingHalvingA protocol event that cuts the rate of new token emissions by half. Halvings are scheduled in advance, happen automatically at fixed intervals, and are a core mechanism for enforcing declining token supply growth over time.Like a savings account where the interest rate is contractually cut in half every four years. You still earn interest, but the rate drops on a known schedule, and the issuer can't change it without breaking the contract.Read more → schedule from the subnet’s creation date. This creates an asymmetry I will return to.

How emissions flow

The network currently emits 0.5 TAO per block (12-second blocks, so 3,600 TAO per day). The first halving has already occurred; original rate was 1 TAO per block.

Emissions reach subnets in two stages.

Stage 1: Injection (every block). The system calculates net TAO flow into each subnet (staked minus unstaked), smooths it with an 87-day exponential moving average with a 30-day half-life, applies an offset so the worst-performing subnets receive zero, and distributes proportionally. Subnets with sustained negative flows get nothing.

Stage 2: Distribution (every tempo, roughly 72 minutes). Each subnet’s emissions split:

The owner share is fixed globally at 18%. Miners are allocated via Yuma Consensus (the subnet’s internal ranking mechanism). Validators and stakers receive proportional to their stake.

The critical insight: emissions are no longer about which subnet has the highest alpha price or the most politically connected validators. They track where capital is actually flowing. Sustained investment earns emissions. Sustained withdrawals earn nothing.

The zero-emission death spiral

This is by design, but worth understanding. When a subnet enters sustained net outflows, it receives zero TAO and zero alpha emissions. The pool stagnates. Slippage increases. Higher slippage encourages more outflows. More outflows ensure continued zero emissions.

It’s a deliberate selection mechanism. Subnets that can’t attract and retain capital die. The network doesn’t subsidise failure. Currently 10 subnets sit on the deregistration list. When a subnet is deregistered, alpha holders can suffer up to 96.875% loss in value through adverse AMM conversion rates.

This is the risk that most dTAO explainers skip. Alpha tokens aren’t stablecoins. They aren’t even typical altcoins with independent markets. They are AMM-denominated exposure to a specific subnet’s economic viability. If the subnet fails, your alpha is worth close to nothing.

The network today

The registration cost is dynamic: it lowers gradually over time and doubles every time a new subnet is created. At $145,000, creating a subnet isn’t casual. Rate-limited to one creation per four days.

Each subnet supports up to 256 neurons and 64 validators.

Notable subnets

Others include Compute Horde (distributed GPUGPUGraphics Processing Unit. Originally designed to render video game graphics, GPUs turned out to be exceptionally good at the massively parallel math that AI models need. Modern AI training and inference runs almost entirely on GPUs.Like a factory with 10,000 workers doing the same simple task in parallel, versus a CPU which is more like 10 workers each doing different complex tasks. AI training involves doing simple math a million times per second on a million numbers, which is exactly what the GPU factory is designed for.Read more → compute), Data Universe (web data collection), and TAOHash (Bitcoin miningProof of WorkThe original blockchain consensus mechanism where miners compete to solve computationally expensive puzzles. The winner proposes the next block and earns the rewards. Proof of Work secures Bitcoin and most pre-2020 chains.Like a lottery that runs every 10 minutes where the tickets cost electricity. Whoever spends the most electricity buying lottery tickets has the best chance of winning that round's prize. Nobody can fake the result because the proof of their work is verifiable by everyone.Read more → coordination). The network is broad but most subnets are early-stage.

I couldn’t verify top subnets by stake ranking. taostats.io loads this data via authenticated APIAPIApplication Programming Interface. A structured way for one piece of software to talk to another. In DeAI, APIs let applications request inference from a model without running the model themselves.Like a waiter in a restaurant. You don't walk into the kitchen and cook your own meal. You tell the waiter what you want, they tell the kitchen, the kitchen cooks it, and the waiter brings it back. The API is the waiter.Read more →. That’s a transparency gap. For a network that prides itself on market-driven allocation, the emission leaderboard should be publicly readable without authentication.

The asymmetric halving problem

A structural issue flagged in GitHub (issue #1975, August 2025) but not resolved.

Alpha emission starts at 1 per block for every new subnet, following an independent halving schedule from the subnet’s creation date. But TAO emission is network-wide and has already halved to 0.5 per block.

Older subnets have lower alpha emissions (their alpha halving is further along) while newer subnets start with full alpha emission. But the TAO injection available to all subnets is already halved at the network level.

The result: early subnets and late subnets face structurally different economics. Early subnets benefited from full TAO emission rates when they were young. Late subnets start with full alpha emission but compete for a smaller TAO pie. Whether this matters in practice depends on pool dynamics, but it’s an asymmetry worth understanding if you’re evaluating subnet participation timing.

What this means for TAO holders

If you hold TAO and aren’t stakingStakingLocking up a cryptocurrency to help secure a blockchain network, usually in exchange for rewards. The locked tokens act as a security deposit that can be taken away if the staker misbehaves.Like putting down a large rental deposit for an apartment. You get the money back if you behave, you earn interest while it's locked, and the landlord takes it if you trash the place.Read more → into subnets, you’re earning nothing and being diluted by 3,600 TAO per day in new emissions. The system rewards active participation, not passive holding.

Your options:

Stake into Subnet 0 (Root). The simplest path. No subnet-specific risk beyond TAO price. You earn a share of the 41% validatorValidatorA computer that runs the full blockchain protocol, verifies transactions, and proposes new blocks. Validators are the workers that keep a Proof of Stake network running, and they earn rewards for doing the work correctly.Like a notary public who witnesses and stamps legal documents. Validators witness transactions, check they follow the rules, and stamp them into the permanent record. A notary who commits fraud loses their license. Validators work the same way, except the license is staked tokens that get slashed on misbehaviour.Read more →/staker allocation from Root’s emissions. Lower returns than successful individual subnets, but lower risk.

Stake into individual subnets. Higher potential returns if you pick subnets with strong net inflows. Higher risk if the subnet enters a death spiral. Your alpha tokens carry subnet-specific risk on top of TAO price risk. Due diligence on subnet utility, team, and flow trends matters.

Provide concentrated liquidity. Highest complexity. Earn swap fees within your specified price range. Requires active management and understanding of impermanent loss dynamics. Not recommended unless you’re experienced with DeFiDeFiDecentralised Finance. Financial services like lending, trading, and yield farming built on smart contracts instead of traditional banks or brokerages. DeFi protocols are usually permissionless and global.Like a vending machine that can give you a loan, swap your currencies, or invest your savings. Nobody is behind the counter, the rules are written into the machine itself, and anyone with money in the right format can use it.Read more → liquidity provision.

Run a subnet. If you have $145,000 in TAO and a genuine AI workload, you can create a subnet and earn the 18% owner share. The bar is high but the economics for successful subnets are compelling. 18% of daily emissions directed to your subnet is meaningful revenue if you can sustain positive net flows.

What I’m watching

1. Flow concentration. How many subnets capture the majority of emissions? If 10 subnets capture 80% of flows, the long tail of 119 subnets is fighting over scraps. The emission distribution curve matters more than the total number of subnets.

2. Deregistration rate. 10 subnets currently on the deregistration list out of 129 active. If this accelerates, it validates the selection mechanism but also means more alpha holders taking losses. The rate of subnet failure is a health indicator in both directions.

3. The asymmetric halving resolution. GitHub issue #1975 has been open since August 2025. How the team addresses (or doesn’t address) this structural unfairness between early and late subnets will signal how seriously they take long-term economic design.

4. Real utility metrics. Which subnets are producing useful AI output versus which are pure staking plays? Compute Horde, Data Universe, and the pretraining subnets should have measurable throughput. If they do, publish it. If they don’t, the subnet economy is speculation on speculation.

5. Registration cost trajectory. At $145,000, subnet creation is restricted to well-funded teams. If TAO appreciates, the dollar cost rises further. This could become a barrier to the subnet diversity the system needs.

The honest assessment

dTAO is genuinely well-designed. Replacing political allocation with market signals is the right direction. The flow-based model (replacing the original price-based model within nine months) shows the team iterates when gaming is identified. The 87-day EMA window, the zero-emission floor for net-negative subnets, the per-subnet halving schedules: thoughtful mechanisms.

But the complexity is real. Understanding alpha tokens, AMM slippage, concentrated liquidity, flow-based emissions, and subnet-specific halving schedules requires more economic literacy than most crypto participants have. The gap between “buy TAO on an exchange” and “make informed subnet staking decisions” is wide.

The system works as designed: capital flows to useful subnets, useless subnets die, and emissions reward genuine investment. Whether the current 129 subnets are mostly useful or mostly speculative is the question I can’t answer from public data. That opacity (the difficulty of assessing individual subnet utility) is the biggest risk for anyone staking beyond Subnet 0.

For our full assessment of Bittensor including the halving analysis, see the Bittensor halving piece and our Bittensor review.