The bullish thesis for TAO requires you to believe in a game-theoretical miracle. But the cryptocurrency industry has seen such miracles before.
Bittensor holds one of the most elegant narratives in the cryptocurrency space: a decentralized AI oracle market where the market mechanism allocates funding to the most impactful research. TAO is the coordination layer, subnets are the labs, and the market is the grant committee.
Stripped of its narrative facade, you'll discover some more unsettling things.
Bittensor is a grant program where cryptocurrency speculators fund AI R&D—and the grantees have no obligation to return any value to TAO.
You can think of TAO as Elon Musk—he is the first donor of OpenAI, a "non-profit" entity. Subnets are like Sam Altman—they are the fund recipients, building and delivering the product, but without any contractual obligation to share the returns. They may ultimately choose to privatize the gains rather than return any value to the initial funders.
Bittensor distributes TAO tokens to subnet operators and miners based on the subnet token price. Once a subnet receives TAO allocation, there is no mandatory mechanism requiring the AI models, datasets, or services produced to stay within the Bittensor ecosystem. Subnet operators can use Bittensor's TAO incentives to rug pull and take the real product elsewhere—onto centralized cloud servers, packaged as a standalone API, or directly sold under an SaaS shell.
TAO has no equity and no authorized contract. The only tether is the subnet token—the token price must hold up to maintain access to resources. But this is only effective before the subnet "flies away": once the product is robust enough to stand alone outside the Bittensor system, this lifeline is cut. The relationship between Bittensor and subnets, rather than being venture capital, is more like research funding—a grant to kickstart you but without taking a share.
To put it bluntly, Bittensor is fundamentally a wealth transfer: from token speculators' pockets to AI researchers' accounts—or more plainly, from lambs to tech-savvy "miners."
The principle is simple: TAO investors act as the backstop for the entire ecosystem. They buy and hold TAO to support the token price, which serves as the channel for funds flowing into the subnetwork incentive system. Subnetwork operators earn TAO inflation rewards through "performance display," which is largely about maintaining a good-looking token price for their subnetwork. The AI products built with this funding can exit at any time—the only constraint being the need to continue acquiring network resources.
This is the VC's worst nightmare: you've invested the money, the product has been built, but they don't owe you anything. All that's left is a token issuance schedule, plus a prayer.
Optimist's Interpretation
Now, let's look at it from a different perspective. The optimistic view is built on two pillars:
The ongoing need for resources means AI companies are always facing a funding shortage. Computing, data, and talent costs are high. If Bittensor can reliably provide these resources at scale, the subnetworks have a reasonable incentive to stay—not because they are locked in, but because leaving would mean losing a channel of resource supply.
There is a soft support in logic: AI's demand for resources is endless, and the scale that TAO can provide is unattainable through self-financing alone. Following this logic, subnetwork teams would proactively maintain their token valuation, without needing any enforcement mechanism, and the TAO economy would naturally form a positive feedback loop. Cryptocurrencies have excelled in aggregating resources. Bitcoin has aggregated massive computing power solely through token incentives. Ethereum's proof-of-work mechanism has also been hugely successful, becoming a powerful magnet for computing resources.
Bittensor is applying the same strategy to the field of artificial intelligence. The "enforcement mechanism" is the token game itself— as long as TAO holds value, the incentive to participate will continue to grow.
If Bittensor's future were simulated 1000 times, the distribution of results would be extremely skewed.
In most simulation scenarios, Bittensor will remain a niche funded project. The AI AI results generated by subnetworks will be insignificant. The best-performing subnetworks will gain significant attention, seize rewards, and then turn to a closed-source mode, providing no value to TAO. When the token issuance exceeds the created value, the TAO token will depreciate.
In a few simulation paths, something really takes off. A certain subnetwork delivers truly competitive AI services, and network effects begin to snowball. TAO becomes the coordination layer of decentralized AI infrastructure in the true sense—not capturing value through enforcement but relying on the gravitational pull inherent in being a reserve asset of a functioning AI economy.
In rare cases, TAO becomes an existence that defines an entirely new asset class.
Where Things Could Go Wrong
The bear case logic is simple: no stickiness. Once a subnet no longer needs TAO token incentives, it will leave. Bittensor is a transitional phase, not a final destination. Centralized AI holds overwhelming dominance. Companies like OpenAI, Google, and Anthropic have orders of magnitude more computing power and talent reserves. TAO cannot compete with the deep pockets of venture capital and private equity markets. Therefore, the best talent will opt for the traditional development path. Inflation is taxation.
TAO's inflation plan subsidizes subnets through dilution of holders. If the value created by the subnet does not match this dilution level, it is a slow bleed disguised as a "growth mechanism."
The optimistic scenario, frankly, looks more like wishful thinking than a realistically viable path to success.
Conclusion
The majority of capital poured into TAO will ultimately subsidize development activities that do not return value to token holders. However, Crypto has repeatedly demonstrated that token incentive-driven coordination games can produce results that defy all rational models.
Bitcoin theoretically should not have succeeded, but it did—even though this argument alone is not sufficient, the industry has used it to endorse numerous projects that cannot withstand first-principles scrutiny.
The core issue with TAO is not whether a coercive mechanism exists—it does not, and the efforts of dTAO have not changed that. The core issue is: whether game theory incentives are strong enough to keep the highest-quality subnets on track. Buying into TAO is a bet that a "soft guarantee" can hold up in a harsh reality.
This is either naive or visionary.
