Author: Max.S

The belief in "Decentralized AI" (DeAI) in the capital markets is facing an unprecedented stress test.

Recently, the leading project in the decentralized AI space, Bittensor ($TAO), experienced a catastrophic internal upheaval. One of the top development teams within the Bittensor ecosystem, Covenant AI, which had just successfully trained a 72B large language model, suddenly announced its complete withdrawal from the Bittensor network via social media. In its departure statement, Covenant AI pointed a finger at Bittensor's founder, Jacob Steeves, harshly criticizing his "absolute and dictatorial" control over the network, accusing him of arbitrarily cutting off token rewards for subnetworks, and bluntly stating that the so-called decentralized AI is merely a carefully orchestrated "charade."

As a result of this black swan event, the price of the $TAO token encountered panic selling in the secondary market, with a single-day drop of as much as 15% to 25%, resulting in a market value evaporation of hundreds of millions of dollars. While the crypto community was "watching the drama" of the public split between the top team and the founder, it also began to seriously examine a deeper industry proposition: in the AI field, which is heavily reliant on computational power capital and complex engineering, is token economics-driven "decentralization" a utopia that reshapes production relationships, or merely a glamorous facade that conceals centralized power?

To understand the destructive power of this event, one must first recognize the significance of Covenant AI within the Bittensor ecosystem.

In Bittensor's multi-subnetwork architecture, most subnets are still at the low stages of API calls, model fine-tuning, or simple task routing, with very few teams capable of training from scratch or conducting large-scale parameter model training. Covenant AI is the "hardcore" representative within this ecosystem. Shortly before its announcement to exit, the team had just delivered a milestone achievement to the community: successfully training an open-source large model with 72 billion parameters (72B) in a decentralized network environment.

Given the current computational costs, training a 72B model means mobilizing a massive GPU cluster (typically equivalent to thousands of H100s running continuously for weeks) and incurring extremely high hardware and electricity costs. The reason Covenant AI was willing to bear such huge upfront sunk costs lies in Bittensor's "Emissions" mechanism— as long as its provided model and computational power achieve high scores in subnetwork evaluations, it can continuously earn $TAO token releases as generous rewards. This is the most attractive flywheel effect in the DeAI narrative.

However, the flywheel came to an abrupt halt at its peak. According to Covenant AI, after investing large sums to complete the training of the 72B model and going live, founder Jacob Steeves and his stakeholders directly cut off the token rewards flowing to the Covenant AI subnetwork by controlling the Validator nodes, without any warning or transparent governance process.

For miners and developers, cutting off Emissions is tantamount to "unplugging the network cable." The ROI on their massive computational expenditures instantly dropped to zero, triggering Covenant AI's furious exit due to such extreme and unpredictable systemic risk.

The term "charade" used in Covenant AI's exit statement accurately struck at the most vulnerable nerve of Bittensor: network control.

The underlying design of Bittensor relies on the Yuma consensus, where "Validators" assess the contributions of "miners" and decide how the system's additional $TAO tokens are distributed. Theoretically, this is a decentralized game system based on the amount of stake and algorithms. However, Covenant AI's accusations reveal a brutal reality: computational power is decentralized, but power and capital are highly centralized.

In the current root network of Bittensor, the leading Validator nodes that can dominate the flow of token distribution have their staking chips highly concentrated among early investors, foundations, and addresses associated with founder Jacob Steeves. This means that the founder is not only the rule maker but also the ultimate judge.

Covenant AI pointed out that when the output of a subnetwork does not align with Jacob's personal wishes or threatens the interests of other "affiliated" subnets, Jacob can easily use his massive staking weight to alter the Yuma consensus distribution results. Such "one-man rule" intervention renders the decentralization at the smart contract level virtually meaningless. Developers spending millions of dollars on computational power find that their fate ultimately depends on one founder's subjective will or opaque manipulations.

Objectively, Jacob and his supporters may argue that they are "maintaining the overall quality of the network" and "preventing specific subnets from exploiting loopholes to mine tokens." However, in the absence of a transparent DAO governance mechanism and no on-chain hearings or appeals process, such "centralized intervention" under the guise of "acting on behalf of the heavens" severely undermines the core value of the network as "a credible neutral infrastructure."

The $TAO token's single-day drop of 15-25% is not merely a panic-induced stampede by retail investors; it is revaluation by institutional funds concerning Bittensor's "governance risk discount."

The reason Bittensor has been able to support a vast market value and enjoy a high valuation premium is that the market views it as the only realistic target for a "decentralized OpenAI." The foundation of this grand narrative is that the system must possess a high degree of predictability: as long as you contribute computational power and high-quality models, the protocol will automatically ensure your returns through code.

The Covenant AI incident shattered this expectation. The top financial practitioners and institutional investors detest "unpredictable single node failures," and here, that failure point is Jacob Steeves's power.

If even the absolute leading team capable of training a 72B model can be rendered helpless due to the founder's intervention, then for other waiting token holders, computational power providers, and AI research institutions, deploying heavy assets on Bittensor is undoubtedly a risky game of Russian roulette that could be "flipped over" at any moment. As high-quality supply sides (miners and developers) refuse to enter due to the fear of centralized tyranny, the application scenarios and inherent value of the $TAO token also become as lifeless as a source without water. The frenzy of capital flight is a preemptive vote against this deterioration of the fundamentals.

Covenant AI's departure is not just a public relations crisis for Bittensor; it is an inevitable growing pain faced by the entire decentralized AI space as it reaches deeper waters. It brutally exposes the "impossible triangle" in the DeAI field: model quality and scale, decentralized credible neutrality, and alignment of incentives to prevent malicious acts.

Centralization of scale vs. decentralization of mechanisms: Training cutting-edge AI (such as models above 72B) is typically a capital-intensive, centralized engineering endeavor that requires highly coordinated GPU clusters. This presents a natural physical divide from the permissionless, decentralized nodes advocated by Web3.

Preventing token manipulation vs. credible neutrality: In order to prevent low-quality nodes from cheating through mutual traffic to siphon off tokens (witch attacks), the network must introduce subjective "quality assessments." However, in today's environment where AI assessment standards have not been fully objectively mathematically defined, once this assessment power is entrusted to a small number of validators, it can easily evolve into centralized power rent-seeking.

Bittensor attempts to bridge the gap between the two with token economics, but the Covenant incident proves that the bridge's supporting pillar (governance mechanism) is currently still weak.

Covenant AI's exit has burst the romantic bubble of Bittensor's "absolute decentralization." For $TAO, this may be a painful moment of disenchantment, but for the entire DeAI industry, it is a necessary wake-up call.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。