Subnet investing

dTAO is one of the most ambitious experiments in crypto.

It turns Bittensor into a market-based allocation system where subnets compete for TAO, emissions, miners, validators, users and investor attention. In theory, this is powerful. Instead of one central committee deciding which AI networks deserve capital, the market continuously reallocates capital toward the subnets it believes are most valuable.

But this also creates a hard and uncomfortable question:

What is a subnet token actually worth?

At the moment, I think the honest answer is: the market does not really know yet.

That does not mean dTAO is broken. It means dTAO is extremely early. We are probably closer to the internet in the 1990s than to a mature financial market. Many subnets will not survive. Some may become very important. Most will be mispriced for long periods of time. And investors need to be careful, because high APY can easily hide extreme price risk.

A subnet price is not the same as a fundamental valuation

Technically, a subnet token has a price because every subnet has its own liquidity pool. The price is determined by the relationship between TAO reserves and alpha reserves.

But that technical explanation does not answer the more important question.

If a subnet trades at 0.057 TAO per token, does that mean the subnet is fundamentally worth 0.057 TAO? Not necessarily. It means that this is where the market currently prices that subnet token based on staking flows, unstaking flows, liquidity, emissions, confidence, speculation and narrative.

Take Affine as an example. If Affine trades at 0.057 TAO, could the subnet still technically function at 0.01 TAO? Probably yes. The code does not stop working because the token is ten times cheaper. Miners can still mine. Validators can still validate. The subnet can still produce outputs.

But the lower price does matter economically.

A lower subnet price can reduce the dollar-denominated value paid to miners and validators. For compute-heavy subnets, that can be a serious issue because GPU costs are real. If miner rewards fall too far, the subnet may struggle to attract high-quality supply. For lighter subnets, the effect may be less dramatic because operating costs are lower.

So the answer is nuanced: a lower subnet price does not automatically break a subnet, but it changes the economic gravity around it.

Chutes: a working subnet, a falling token

Chutes is one of the most important case studies in dTAO.

It is not an empty narrative subnet. It has real usage, real revenue, serious infrastructure and a visible buyback model. It is one of the clearest examples of what many people hope dTAO subnets will become: useful digital infrastructure with revenue that supports the subnet token.

And yet, the token has been destroyed in price terms.

Chutes traded around 0.4 TAO per token in 2025. More recently, it has traded around 0.067 TAO. That is a decline of more than 80% in TAO terms.

For an investor, that is brutal.

Imagine staking 1 TAO into Chutes near 0.4 TAO per token. You would receive roughly 2.5 Chutes tokens. If the token later trades around 0.067 TAO, those 2.5 tokens are worth only around 0.1675 TAO before accounting for staking rewards. That means your position has lost more than 80% in TAO terms not considering staking rewards.

This is the key lesson: even a serious subnet can produce terrible investor returns if the entry price is too high.

That does not mean Chutes is bad. In fact, the opposite may be true. Chutes may be making real progress as a product and as infrastructure. But the token price can still fall if the market previously overpaid, if emissions create too much sell pressure, if broader sentiment weakens, or if revenue is still too small relative to the token valuation.

This is one of the most important points for dTAO investors to understand:

A good subnet is not automatically a good investment at every price.

Buybacks matter, but they are not magic

The theory behind buybacks is simple.

If a subnet generates revenue and uses that revenue to buy back its own token, it creates structural buy pressure. This can partly offset miner sell pressure. It can make the token scarcer. It can align subnet success with token holders. And if the buyback pressure becomes large enough, it may help stabilize the token price.

This is the optimistic dTAO model.

A subnet generates useful work. Users pay for that work. The subnet uses part of the revenue to buy back alpha. Investors notice. More TAO flows into the subnet. The subnet receives more emissions. Miners earn more. The service improves. More users come in. More revenue is generated. More buybacks happen.

That is the flywheel.

But the current market does not always price this correctly.

Chutes has buybacks and revenue (though not enough to offset miner sell pressure), yet its token has fallen heavily. Bitcast has shown a clear revenue and buyback policy, yet its token has also seen serious price weakness. Yanez and RedTeam reportedly perform daily buybacks and trade around extremely low levels, around 0.004 to 0.006 TAO. Meanwhile, other subnets without obvious revenue or buyback discipline can trade at much higher prices.

This creates a very uncomfortable question for revenue-generating subnets:

Why do buybacks if the market does not reward them?

If a subnet like Yanez or RedTeam spends real money every month buying back its token, while many higher-priced subnets generate no revenue and perform no buybacks, the rational frustration is obvious. From the team’s perspective, it may feel like throwing money into a market that is not paying attention.

But this is probably a temporary feature of an immature market.

In the short run, prices are driven by narratives, founder reputation, category excitement, trader rotation and speculation. In the long run, if dTAO matures, the market should become more demanding. It should begin to distinguish between subnets with real revenue and subnets with only promises. It should begin to value transparent buyback policies. It should begin to punish subnets that extract emissions without returning value to token holders.

The question is not whether buybacks matter.

The question is how long it takes before the market starts caring.

The market currently prices many subnets almost randomly

This is my biggest concern with the current dTAO market.

There are research-based subnets with high valuations. There are revenue-generating subnets with very low valuations. There are agent subnets with extremely high prices, partly because they are associated with important people in the Bittensor ecosystem. There are subnets with almost no visible revenue trading higher than subnets that are actually buying back tokens every day.

This does not mean the market is stupid. It means the market is young.

In early markets, narrative often leads fundamentals. A subnet connected to a powerful future category, such as agents, decentralized inference or AI research, may trade at a premium even before revenue appears. Investors are not only pricing current cash flows. They are pricing possibility.

That can be rational to some extent.

A subnet like Minos may have no meaningful current revenue but still trade higher than a revenue-generating subnet because the market believes its future potential is larger. An agent subnet like Arbos may trade at a high price because investors believe agent infrastructure could become one of Bittensor’s most important categories, especially if it is associated with highly respected ecosystem figures.

But there is a fine line between pricing future potential and pricing air.

Right now, much of dTAO sits on that line.

This is why investors should be careful with simple stories. “This subnet has high APY” is not enough. “This subnet has a famous founder” is not enough. “This subnet is in an exciting category” is not enough. “This subnet does buybacks” is also not enough by itself.

The real question is whether the subnet can create durable value, defend its position, attract users, align token holders and survive long enough for the market to recognize that value.

What does a subnet token actually represent?

This is the central question behind dTAO.

A subnet token is not equity. It does not give legal ownership of the company behind the subnet. It does not give a contractual claim on revenue. It does not automatically entitle the holder to cash flows.

Instead, a subnet token represents exposure to a subnet’s position inside Bittensor’s incentive system.

That position can become valuable if the subnet attracts TAO, receives emissions, builds useful infrastructure, generates revenue, supports its token through buybacks, and maintains investor confidence.

But none of this is guaranteed.

This makes subnet tokens very different from traditional financial assets. They are not clean equity claims. They are not simple yield instruments. They are liquid coordination assets in a live economic game.

That is both the beauty and the danger of dTAO.

High APY can destroy investors

One of the most dangerous mistakes in dTAO is treating APY as if it were the same as return.

It is not.

If a subnet offers very high APY but its token falls by 50%, 70% or 90%, the investor can still lose badly in TAO terms. This is especially important because many investors look at subnet staking dashboards and focus on the yield number without properly considering token price risk.

This is how people get hurt.

They see a subnet with high APY. They stake TAO into it. They receive more alpha tokens. But then the alpha token falls so much that the yield does not matter.

This is why dTAO investors should think less like passive yield farmers and more like early-stage venture investors operating in a liquid market.

The question is not: where is the highest APY?

The question is: what is the probability that this subnet token preserves or increases its TAO-denominated value over time?

That is a much harder question.

The lower bound is also unclear

A few months ago, some of the lowest-ranked subnets traded around 0.001 TAO. More recently, some low-end prices appear closer to 0.003 TAO.

But what is this floor based on?

Is it based on fundamental value? Probably not.

It is more likely based on a mix of speculation, registration cost, optionality, emissions, market psychology and the idea that even a weak subnet may have some value as a live slot inside Bittensor.

But again, this is not a true valuation model. It is price discovery in a very young market.

That matters because investors may mistake a temporary market floor for a fundamental floor. A subnet trading at 0.003 TAO may look “cheap,” but cheap compared to what? If there is no revenue, no users, weak team execution and no buyback policy, it may still be expensive.

In dTAO, low price does not automatically mean good value.

High price does not automatically mean quality either.

The game theory of buybacks

In the long run, I expect buybacks and revenue transparency to matter much more than they do today.

Imagine two similar compute subnets. Both provide useful GPU-based services. Both compete for miners, validators and investors. One uses 50% of revenue for transparent buybacks. The other uses nothing or only 10%.

As an investor, why would I stake into the second subnet if the first one is clearly more aligned with token holders?

Over time, this should create competitive pressure. Subnets that generate revenue and return part of that revenue to the token should become more attractive. Subnets that generate revenue but do nothing for token holders may be punished. Subnets that generate no revenue and have no credible path to revenue may eventually lose market confidence.

This is the positive long-term vision for dTAO.

Bittensor becomes a giant evolutionary machine. Subnets compete not only on technical performance, but also on economic alignment. The market begins to filter. Weak subnets lose capital. Strong subnets attract it. Revenue-producing subnets gain power. Buyback discipline becomes a competitive advantage.

If this happens, the demand for TAO could become much larger.

Why? Because to gain exposure to subnet tokens, investors need TAO. If high-quality subnets increasingly perform buybacks, and if investors increasingly allocate toward revenue-generating subnets, then TAO becomes the gateway asset into a growing internal economy.

That is the bullish theory.

But it depends on one crucial assumption: the market must become better at recognizing quality.

Right now, I am not convinced it does.

Why transparent buyback policies matter

One of the most practical improvements would be standardized buyback and revenue reporting.

Every subnet should make it easy for investors to answer basic questions:

How much revenue does the subnet generate?

How much of that revenue is used for buybacks?

Are buybacks automatic or discretionary?

How often do buybacks happen?

Are bought-back tokens burned, restaked or held in a treasury?

How much miner sell pressure exists?

What percentage of emissions is covered by revenue?

What is the subnet’s cost structure?

Without this information, investors are flying almost blind.

Subnets like Chutes and Bitcast already provide more clarity than most. That is good. But this should become the standard, not the exception.

If dTAO wants mature capital, it needs mature reporting.

Not because every subnet should be forced into the same model. Research subnets, compute subnets, agent subnets, data subnets and consumer-facing subnets will all have different paths to value. But investors need comparable information to judge whether a subnet is making progress or simply absorbing emissions.

Root APY and the capital allocation problem

There is also a broader question around root APY.

Some people have suggested that root APY should be reduced or removed so that TAO holders are pushed more strongly into subnets. The logic is understandable. If holders can earn relatively safe yield on root, they may avoid subnet risk. Lowering root yield could increase subnet participation and deepen dTAO markets.

But this is controversial for a good reason.

A conservative root yield may be healthy. It gives long-term TAO holders a base option. It allows people to believe in Bittensor without forcing them to underwrite individual subnets. It may reduce reckless behavior and make the ecosystem more attractive to cautious investors. It is like buying index funds such as the S&P500.

Forcing everyone further out on the risk curve (like picking individual stocks) could increase subnet liquidity, but it could also make the system feel less stable.

The better question is not simply whether root APY should be removed. The better question is what balance between conservative TAO holding and active subnet risk-taking best supports Bittensor over the long run.

How dTAO may evolve

I think dTAO will probably go through several phases.

The first phase is speculation. This is where we are now. Prices move heavily on narrative, category excitement, founder reputation, APY and trader flows. Many subnets are mispriced. Some serious teams are undervalued. Some weak subnets are overvalued. The market is noisy.

The second phase is differentiation. Investors begin to separate subnets by category. Compute subnets are judged differently from research subnets. Agent subnets are judged differently from storage subnets. Revenue-generating subnets are separated from purely experimental ones.

The third phase is transparency. Subnets publish revenue, buybacks, emissions coverage and treasury policies. Investors begin to reward clear alignment and punish opacity.

The fourth phase is consolidation. Weak subnets disappear, get deregistered or lose relevance. Stronger subnets accumulate more capital, more miners, more users and more trust.

The fifth phase is institutional understanding. If Bittensor succeeds, outside investors eventually stop asking “which subnet has the highest APY?” and start asking more serious questions: which subnet has durable demand, defensible infrastructure, credible revenue and tokenholder alignment?

That is when dTAO becomes more investable.

We are not there yet.

The real lesson for investors

The main lesson is simple:

dTAO is early. Very early.

This is not a mature market with stable valuation models. It is a live experiment in decentralized capital allocation. The upside could be enormous if Bittensor succeeds in creating a market for machine intelligence and digital services. But the risk is also very real.

Investors should not assume that high APY protects them. They should not assume that buybacks automatically create a floor. They should not assume that a famous founder, an exciting category or a high subnet rank guarantees future returns.

They should ask harder questions.

What does this subnet produce?

Who uses it?

Does it generate revenue?

Does that revenue flow back to the token?

How much sell pressure exists?

Are miners being paid enough to keep producing quality work?

Is the team transparent?

Does the token have any reason to appreciate besides speculation?

And most importantly: am I buying this subnet at a price that already assumes success?

That last question matters most.

Because in dTAO, as Chutes shows, even a strong subnet can be a bad investment if bought at the wrong price.

Conclusion: dTAO is not broken, but it is immature

The point of this article is not to attack dTAO.

I think dTAO is one of the most interesting experiments in crypto. It may become a new way to fund, coordinate and scale digital infrastructure. It may force subnets to become economically honest. It may turn TAO into the gateway asset for a large internal economy of AI, compute, data, agents, storage and other digital commodities.

But today, the market is still learning.

It does not yet know how to value subnets properly. It does not consistently reward revenue. It does not consistently reward buybacks. It often rewards narrative more than fundamentals. It sometimes prices possibility higher than actual performance.

That is normal in a young market.

But investors need to understand what kind of game they are playing.

dTAO is not passive yield. It is not simple staking. It is not equity investing. It is not venture capital either, although it has similarities. It is a new kind of liquid, reflexive, incentive-driven capital market.

That makes it exciting.

It also makes it dangerous.

The next few years will likely determine whether dTAO becomes a serious capital allocation machine or just another speculative rotation game. My guess is that it will become more serious over time, but only if the market learns to reward the right things: real usage, real revenue, transparent buybacks, credible teams, aligned incentives and long-term utility.