TAO is Elon Musk, who invested in OpenAI, and Subnet is Sam Altman

Author: Momir, IOSG

The bullish logic of TAO requires you to believe that a game theory miracle can occur. However, such miracles have happened in the cryptocurrency industry before.

Bittensor has one of the most elegant narratives in the cryptocurrency space: a decentralized AI intelligence market where the market mechanism allocates funds to the most influential research. TAO is the coordination layer, the subnet is the laboratory, and the market is the funding committee.

Peeling away the narrative, you will find something more disturbing.

Bittensor is a funding program where cryptocurrency speculators provide funding for AI research—while the recipients have no obligation to return any value to TAO.

You can think of TAO as Elon Musk—he was the first investor in OpenAI, a "non-profit" organization. The subnets are like Sam Altman—they are the builders who receive funding and deliver products, but have no contractual obligation to share profits. They may ultimately choose to privatize the profits without returning any value to the original source of funding.

Bittensor will distribute TAO tokens to subnet operators and miners based on the price of subnet tokens. Once a subnet receives a TAO allocation, there is no enforcement mechanism requiring that the AI models, datasets, or services they generate must remain within the Bittensor ecosystem. Subnet operators can use Bittensor's TAO incentives to exploit the system and then move the actual products elsewhere—onto centralized cloud servers, packaged as independent APIs, or directly sold under a SaaS shell.

TAO has no equity and no licensing contracts. The only binding factor is the subnet token—the token price must hold up to maintain access to resources. But this is only effective before the subnet has "flown away": once the product is robust enough to stand on its own outside the Bittensor system, this tether is severed. The relationship between Bittensor and the subnet resembles venture capital less than it does research funding—providing you with startup capital but not taking any of your equity.

To put it bluntly, Bittensor is essentially a wealth transfer: from the pockets of token speculators to the accounts of AI researchers—or more straightforwardly, from the inexperienced to the "miners" who understand the technology.

The principle is simple:

• TAO investors are footing the bill for the entire ecosystem. They buy and hold TAO to support the token price, and the token price itself is the conduit for funds flowing into the subnet incentive system.

• Subnet operators earn TAO inflation rewards by "showing performance"—but in reality, "showing performance" largely means keeping their subnet token price looking good.

• The AI products built with this funding can leave at any time—the only constraint is that they still need to continuously acquire network resources.

This is the nightmare that VCs fear the most: you put in the money, they produce the product, but they owe you nothing. What remains is a token issuance timetable, plus a prayer.

I. Optimistic Interpretation

Now let's look at it from a different angle. The optimistic view is built on two pillars:

1. Continuous resource demand leads AI companies to always face funding shortages. The costs of computing, data, and talent are high. If Bittensor can reliably provide these resources at scale, subnets will have a reasonable incentive to stay— not because they are locked in, but because leaving means losing access to resource supply channels. There is a soft logical support: the demand for resources in AI is endless, and the scale that TAO can provide is unattainable through self-funding alone. Following this logic, subnet teams will actively maintain their token valuations without needing any enforcement mechanisms, allowing the TAO economy to spontaneously form a positive feedback loop.

2. Cryptocurrency excels at resource aggregation. Bitcoin has aggregated massive computing power solely through token incentives. Ethereum's proof-of-work mechanism has also achieved great success, becoming a powerful magnet for computing resources. Bittensor is applying the same strategy to the AI field. The "enforcement mechanism" is the token game itself—as long as TAO has value, the motivation to participate will continue to grow.

If we simulate the future of Bittensor 1000 times, the distribution of outcomes will be extremely skewed.

In most simulations, Bittensor will remain a niche funding project. The AI achievements produced by the subnets will be insignificant. The best-performing subnets will gain significant attention, seize rewards, and then shift to a closed-source model, leaving no value for TAO. When the token issuance exceeds the value created, TAO tokens will depreciate.

In a few simulation paths, something truly takes off. A subnet creates a genuinely competitive AI service, and network effects begin to snowball. TAO becomes the true coordination layer of decentralized AI infrastructure—not capturing value through coercive constraints, but through the gravitational pull of being a reserve asset in an operational AI economy.

In very rare cases, TAO becomes the existence that defines a whole new asset class.

II. Where Problems May Arise

The bearish 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 an overwhelming advantage. Companies like OpenAI, Google, and Anthropic possess orders of magnitude more computing power and talent reserves. TAO cannot compete with the substantial strength of venture capital and private equity markets. Therefore, the best talent will choose traditional development paths.

• Issuance is taxation. TAO's issuance plan subsidizes subnets by diluting holders. If the value created by the subnet does not match this level of dilution, it is a chronic bleed disguised as a "growth mechanism."

The optimistic scenario, to put it bluntly, is more like wishful thinking than a genuinely feasible path to success.

III. Conclusion

Most of the capital invested in TAO will ultimately subsidize development activities that do not return value to token holders. However, Crypto has repeatedly proven that token incentive-driven coordination games can produce outcomes that all rational models cannot predict. Bitcoin should not have succeeded, yet it did—though this argument itself is not sufficient, the industry has used it to endorse many projects that do not withstand scrutiny from first principles.

The core issue with TAO is not whether enforcement mechanisms exist—they do not, and the efforts of dTAO have not changed this. The core issue is whether the game-theoretic incentives are strong enough to keep the highest quality subnets on track. Buying TAO is betting that a "soft guarantee" can hold up in a harsh reality.

This is either naive or visionary.