Original title: Dynamic TAO: Your No-Nonsense Guide Original author: Yau Teng Yan, founder of Chain of Thought Original translation: ChatGPT

Editor’s note: The article points out that the launch of dTAO addresses the unfair and concentrated reward distribution of Bittensor in the early stages, linking subnet TAO rewards to demand through market-driven mechanisms, directing more rewards to high-demand subnets. This change alters the previously validator-led reward distribution to market pricing through Alpha tokens, ensuring better-performing subnets receive more rewards and incentivizing activity and innovation across the network. Although challenges with manipulation and liquidity remain, dTAO has made significant strides toward decentralization and market-driven approaches.

Below is the original content (for readability, the original content has been reorganized):

A year ago, I wrote an article about the economic challenges of Bittensor. Now, with the launch of dTAO, the network is finally starting to address these issues.

This transformation was long overdue. The growth of Bittensor has outpaced the ability of root validators to fairly assess new subnets. Influence is concentrated at the top, and sometimes this influence is not entirely neutral. dTAO resolves this issue by letting demand, rather than gatekeepers, determine which subnets can thrive.

Honestly, it wasn't easy to understand dTAO at first. It took a considerable amount of time to piece together all the details.

This is also why I wrote this article: to break it down in simple language, avoiding the confusion I felt at the time, and to summarize my main takeaways.

Why dTAO?

First, it's worth asking: what problem is dTAO actually designed to fix?

Validator bottleneck

Under the old system, newly minted TAO was distributed based on validator votes. Theoretically, this 'validator democracy' makes sense, but in practice, it had issues.

As more subnets come online, top validators can hardly keep up. There are too many subnets, signals are too mixed, and bandwidth is insufficient to fairly evaluate them.

Over time, the system fell into a state of apathy — a few well-connected subnets received most of the rewards, while new subnets struggled to gain support.

Conflicts of interest

Many top validators happen to be subnet owners. By controlling the distribution of rewards, some allocate higher weights to their own subnets, thereby effectively increasing their own rewards.

Some even described the previous system as more like 'giving TAO to their friends'.

In extreme cases, validators and subnet owners may reach private agreements or revenue-sharing arrangements, creating centralization risks that distort reward distribution and undermine trust in the system.

Poor coordination

Validators have their own weight distribution methods to determine rewards, but these methods do not always align with the overall goals of the network.

Worse still, stakers providing economic weight cannot directly decide which subnets can receive rewards. Most staked TAO is concentrated in a few validators, who effectively control the entire landscape of reward distribution.

dTAO - Significant Changes

Before the launch of dTAO, reward distribution was very straightforward. Each block produced 1 TAO (i.e., 7,200 TAO per day), and these TAO were allocated to subnets based on weights assigned by root validators. Each subnet allocated TAO to its contributors at a fixed ratio: 18% to subnet owners, 41% to miners, and 41% to validators.

For stakers, everything revolves around Root (Subnet 0). TAO holders stake behind validators, and in return, they receive newly minted TAO. This process is direct and predictable.

But all of that is now over.

Under dTAO, rewards are no longer distributed based on validator weights. Instead, TAO rewards flow through a market-driven system, using subnet tokens, also known as Alpha tokens.

The prices of these Alpha tokens reflect the market demand for a given subnet. The higher the demand, the higher the price, indicating that the subnet is more valuable.

Subnet tokens use a constant product automated market maker (AMM) — the same pricing mechanism used by Uniswap. The price at any given moment is determined by the ratio of TAO reserves to Alpha token reserves in the subnet liquidity pool.

So, rewards are no longer arbitrarily distributed; the TAO reward share for a given subnet is now determined by how the market perceives its value.

· Higher Alpha token prices → Reflect strong market demand → More TAO rewards

· Lower Alpha token prices → Reflect weak market demand → Fewer TAO rewards

Imagine two subnets competing for rewards:

· Token price of Subnet A = $100

· Token price of Subnet B = $50

Due to Subnet A's higher token price, it receives a larger share of TAO rewards per block than Subnet B.

This mechanism creates a self-regulating system where the market believes subnets that create more value (perform better) will receive more TAO rewards, while poorly performing subnets will naturally see lower rewards.

This mechanism ensures that capital flows to the most productive subnets.

Reward distribution

It took me some time to piece together the actual situation of reward distribution, so let me try to summarize it concisely. The chart below is worth referencing multiple times in the coming articles. I find it to be one of the most informative visualizations, showing how the ratio of TAO rewards distributed to Root versus subnets changes over time.

What happened to the Root network?

In the early stages of dTAO (like now), stakers on Root (Subnet 0) will still receive most of the TAO rewards. The reason is simple: the circulation of Alpha tokens is still low.

The staking weight of validators in each subnet depends not only on the share of Alpha tokens they hold in that subnet but also on the amount of TAO they stake in Root.

Because Alpha tokens are still scarce in the early stages, Root validators hold a disproportionate amount of influence across the network.

This led to most of the newly minted Alpha tokens being automatically 'sold back' to TAO and distributed to TAO stakers on Root.

Currently, this is very favorable for Root stakers. Since most rewards still flow into Root, the returns are very generous, with annual percentage rates (APR) reaching 60-70% just a few days ago.

But this won't last long.

As more Alpha tokens enter circulation, their weight in the network will also increase, gradually shifting influence away from Root validators. Over time, dTAO ensures that rewards are no longer concentrated in Root, but are more evenly distributed across the network.

About 100 days later, the TAO rewards received by Root stakers are expected to balance with the subnet Alpha tokens.

What happened to the subnets?

Subnets no longer directly allocate newly minted TAO to miners and validators, but instead use Alpha tokens — their own native currency to reward participants.

How Alpha tokens work

Each block, each subnet will mint Alpha tokens, starting with 2 Alpha from each block (which is twice the TAO minting rate). The minting rate is dynamic; the higher the price of Alpha tokens, the lower the minting rate.

Like TAO, each Alpha token has a hard cap on total supply, set at 21 million tokens, and follows the same halving schedule, with the first halving occurring when total supply reaches 10.5 million — expected to happen within two years.

Each subnet operates an AMM pool, where TAO is paired with Alpha tokens. The price of Alpha is dynamically determined based on the ratio of TAO to Alpha tokens in the pool.

When you stake TAO into a subnet pool, you receive an equivalent amount of Alpha tokens at the current market price. Later, if you unstake, you can exchange the Alpha tokens back to TAO at market price.

When people say they are 'buying' or 'selling' subnet tokens, they are actually just staking or unstaking TAO into the subnet pool. There isn't yet a standalone mechanism for directly purchasing Alpha tokens.

Reward distribution methods

At each block, the protocol scans the Alpha token prices of all subnets and dynamically decides how much newly minted TAO to inject into each subnet's pool.

· Higher-priced subnets (indicating stronger demand and utility) will receive more TAO rewards.

· Lower-priced subnets receive fewer TAO rewards.

Subnets must now earn TAO rewards by generating actual demand for their Alpha tokens. This creates a competitive, market-driven environment where success must be earned through effort, and only the fittest subnets will survive.

Subnets can no longer earn rewards by going through the motions; they must prove their value.

Those subnets that demonstrate real utility and maintain higher Alpha token prices will receive more TAO rewards; while poorly performing subnets will naturally decline, as their rewards will gradually dwindle.

To prevent short-term price fluctuations from distorting reward distribution, the system employs an Exponential Moving Average (EMA) approach, which smooths volatility and ensures that even in the event of sudden price shifts, TAO reward distribution remains stable. Thus, changes in Alpha prices lag in reflecting on TAO reward distribution — usually taking days rather than hours.

Timeline

Here’s a breakdown of key stages to help you anticipate future developments. Note that these are just rough timeframes and conceptual milestones, not set in stone.

Launch phase (Day 0 - Day 1)

· Each subnet mints 2 Alpha per block (starting new minting).

· Since Alpha tokens have not yet circulated, Root will receive about 100% of TAO rewards.

· The staking weight of validators is entirely based on TAO.

· The highest returns for TAO stakers are on Root.

Early adoption phase (Day 2 - Day 30): We are currently in this phase

· As subnets begin minting Alpha, more Alpha tokens enter circulation.

· Validators in subnets begin to accumulate Alpha, with staking weights gradually shifting towards the subnets.

· Root still receives the majority of TAO rewards, but its dominance is starting to decline.

Transition phase (Day 30 - Day 100)

· The supply of Alpha in subnets has rapidly increased due to high minting rates.

· Subnet validators with more Alpha begin to surpass TAO-based Root validators in staking weights.

· The TAO reward share for Root has significantly declined.

· By Day 100, the staking weights of TAO and Alpha will reach a balance, meaning that TAO staking will no longer dominate the validator weight calculations.

Post-transition phase (Day 100 - Year 1)

· Subnet validators dominate reward distribution.

· Root continues to receive TAO rewards, but the rate of rewards has significantly slowed.

· Validators who adapted to staking Alpha tokens in subnets earn the most.

Long term (Year 1 and beyond)

· TAO rewards are almost entirely determined by market-driven subnet staking.

· Root still allows TAO staking, but rewards are minimal, with TAO's weight approaching 0.

· The network is fully decentralized in terms of reward distribution.

What has happened so far

All subnet token pools initially only had 1 Alpha/1 TAO, which can be considered a fair starting model.

Shortly after the launch of dTAO, the price of Alpha experienced wild fluctuations. Some subnets skyrocketed to 5-10 TAO/Alpha within the first few hours, primarily driven by speculators hoping to profit from high rewards. Other subnets maintained prices between 0.1-0.2 TAO, possibly due to the influence of major brands or a lack of early marketing.

But the early price surge may not last long for the following reasons:

· Automatic selling of Root rewards: Many of the newly minted Alpha tokens will automatically sell back into these subnet pools to pay the TAO stakers on Root. This exerts downward pressure on the price.

· Selling pressure: Miners, validators, and subnet owners earn Alpha tokens (which mint faster than TAO) and then exchange them back to TAO to cover operational costs.

Warning: In the early stages, the relative inflation of subnet Alpha tokens is very high. Low circulation, high volatility in price.

The best place I found to track subnet Alpha prices is @BackpropFinance.

As of now, the circulation of Alpha in each subnet is less than 0.4%, so you will often see low market values but fully diluted valuations that are very high, typically reaching billions of dollars.

For example, the recent hottest one is Chutes, which provides serverless GPU resources by @rayon_labs (Subnet 64). It shows a market value of about $9.2 million, but a fully diluted valuation of $2.6 billion, which is quite astonishing compared to the market value of TAO at around $4 billion.

Given the severe inflation and ongoing selling pressure, I expect these fully diluted valuations (FDV) to eventually revert to more realistic levels.

A trackable metric is the total FDV of all subnet tokens compared to the market value of TAO (see the leftmost chart above). Currently, it's around 2-3 times, but clearly, this situation is unsustainable in the long term.

However, this is still a low liquidity environment, meaning that even small-scale 'blind' purchases or sell-offs by traders can quickly drive price fluctuations. Just in the past day, we've seen subnet token prices skyrocket by 100-200%, only to rapidly cool down.

How to acquire subnet Alpha tokens

Choosing the right subnet is the first step.

Not all subnets are the same, so look for those with real utility, active communities, and strong miner participation — tools like Discord, X, GitHub can help you discover genuine market momentum.

Once you find a subnet you like, you need a Bittensor wallet. A good wallet is the official Chrome extension from @OpenTensor Foundation. You can only use TAO staking to acquire subnet tokens, so there’s currently no direct mechanism to swap them for ETH, USDC, or SOL.

While you can stake directly through your wallet, I prefer to use platforms like Taostats.io or Backprop Finance, which offer a familiar trading interface. Just be cautious of slippage: many subnets still have low liquidity, so larger trades can cause significant price fluctuations.

One benefit? Holding Alpha tokens allows you to earn more tokens, as newly minted Alpha tokens will automatically accumulate in your balance. When you want to exit, you can unstake and exchange back to TAO at the rate in the pool. This rate may differ from when you staked, so you could actually incur a loss in TAO.

My thoughts on dTAO

The coolest thing about dTAO and Bittensor subnet tokens is that it makes everyone genuinely care about the construction of each subnet. There's real Alpha — by doing research, you can actually gain upside opportunities — whereas before, you only needed to buy TAO.

And that's precisely the point.

I am taking the time to deeply understand what these subnets are doing, grasping their product visions and business models.

And this is actually very interesting. Many subnets are led by technical teams with deep expertise in AI, working on scientific research ranging from AI model development to protein folding and advanced visual models.

These subnets are at different stages of development. Some subnets are still in the early stages, focusing on building miner communities, while others have begun generating valuable outputs and securing business partnerships.

If I were to give any advice regarding subnet tokens, it would be: patience may be your best strategy.

dTAO is a system designed for the long term, gradually transitioning to reward those who take the time to understand it in depth. If you like a particular subnet, gradually buying in through small periodic purchases (rather than a large one-time investment) can help you weather the volatility while allowing liquidity to grow steadily.

For those who are not fond of the subnet hype, you could likely continue to stake your TAO on Root over the next few months. The APR there will gradually decrease over time, but it remains respectable.

Of course, some may want to play the volatility 'lottery', and if they can accurately time their entry and exit, they may significantly increase their TAO holdings. However, this is a very risky game.

Some thoughts

dTAO is an important step in the right direction. But it is not a perfect system (is there a perfect system?).

Manipulation can still occur

Switching to an Alpha-based staking model introduces a new set of risks. Theoretically, it creates better market-driven incentives, but in practice, it still leaves room for exploitation. If a subnet's token price crashes, malicious actors might seize the opportunity to buy Alpha tokens at a low price and use them to manipulate rewards.

Currently, dTAO mitigates this issue by integrating Root staking weights, making it harder for any single entity to hijack rewards. However, as Root weights gradually decline, subnets will need stronger security mechanisms to prevent hostile takeovers. Subnet owners can still reach private agreements with large validators or miners — the medium now is Alpha, not TAO. While the system is more decentralized, it still cannot avoid the possibility of game-theoretic collusion.

1,000 subnets?

Every time a new subnet is registered, the fee doubles. However, over time, the price will also decay, halving approximately every 38,880 blocks — roughly every five and a half days. This means a new subnet could potentially be launched every five days if demand remains stable.

What happens when we have a thousand subnets?

At such a large scale, no individual can reasonably track all the rewards, performance, and opportunities across the entire network. The volume of data will be too large, with too many variables and too much noise. AI-based analytical tools will become necessities rather than luxuries, helping stakers find direction among overwhelming options.

For those launching new subnets, early liquidity will be key. They need to create an initial wave of demand for their Alpha tokens, or else their reward share will remain insignificant.

Over time, the real winners will be those subnets that demonstrate true practicality, establishing a strong correlation between usage and Alpha prices. Subnets that fail to attract usage may gradually fade away, with their tokens drifting toward zero.

DeFi on Bittensor

I'm excited to see future upgrades potentially bring DeFi-like mechanisms into the ecosystem, further enhancing capital efficiency.

One possibility is to adopt Uniswap V3-style liquidity pools, allowing for concentrated liquidity rather than a constant product AMM model. Another possibility is permissionless liquidity provision, where external LPs can deposit funds into the Alpha–TAO pair and earn trading fees.

These currently do not exist — but if implemented, they could fundamentally change the game, making subnet tokens more attractive for trading and staking.

More capital → More liquidity → More capital.

dTAO is still in its early stages, but with the introduction of new mechanisms, the financial aspect of Bittensor may evolve into something more complex than what we see today.

Bittensor Revival

Last month, in my personal prediction for 2025, I wrote that Bittensor might experience a revival:

Bittensor spent most of last year in a predicament — as the 'veteran' of crypto AI, it was overlooked by emerging narratives. Despite the wave of AI tokens and agents, TAO struggled to keep up.

Now, with the launch of dTAO, this situation is changing. Interest is surging, and for the first time, the network is forcing everyone to pay attention — not just to TAO, but to what each subnet is actually building.

Congratulations to the const and Opentensor teams, as well as everyone who worked hard for all of this. I admit I had my doubts about it.

This is not just an upgrade. It is the beginning of the Bittensor revival, a process that will unfold over many years.

Related resources

· Dynamic TAO White Paper (Warning: There is a lot of math)

· Opentensor's dynamic TAO FAQ

· Taostats (a good resource for tracking network and subnet data)

· Taopill (an overview of the functions and key achievements of each subnet)

· Backprop finance for monitoring subnet token prices

· Wombo's automated subnet evaluation paper

· Many good tweets from @bloomberg_seth, @Old_Samster (@CrucibleLabs), and @xavi3rlu (Latent Holdings)

· @taotimesdotai / @brodydotai: Best communications for Bittensor

· @TAOTalkPod: Excellent podcast on Bittensor

‘Original Link’