Original | Odaily Planet Daily (@OdailyChina)
Author | Azuma (@azuma_eth)
Remember the story of NVIDIA CEO Jensen Huang praising Bittensor (TAO)?
On March 20, during an appearance on Chamath Palihapitiya's All-In podcast, Huang was asked whether he was "optimistic about decentralized AI systems/computing power networks." Palihapitiya cited Bittensor as an example (with a hint of self-promotion), mentioning that a certain subnet team on Bittensor had successfully trained a 4-billion-parameter (actually 72 billion parameters) Llama model, with the entire process completed through distributed computing collaboration. Huang's response was that it was "a remarkable technical achievement."
Boosted by this positive news, TAO surged against the market trend last month, briefly exceeding $370, and Bittensor was seen as "the hope of the entire village" in the cryptocurrency industry.
However, just half a month later, the situation took a sharp turn due to a sudden announcement — as of the morning of April 10, TAO had fallen below $290, declining sharply for three consecutive days, and Bittensor found itself embroiled in what may be its biggest public controversy since its inception.
What Huang Praised Was Actually a Subnet Team Called Covenant AI
Before explaining the details of the incident, we need to first understand Bittensor's subnet architecture.
Bittensor is a decentralized machine learning network centered around token incentives. Through its subnet mechanism, Bittensor allows different teams to build various AI task markets, with miners and validators participating in computation and evaluation to distribute TAO rewards.
The "certain subnet team" mentioned by Palihapitiya is actually called Covenant AI (formerly known as Templar), and the model praised by Huang is called Covenant-72B. This is a model with 72 billion parameters, collaboratively trained in a permissionless manner by over 70 independent contributors on general-purpose hardware, making it the largest decentralized large-scale model pre-training project in history.
In simple terms, Bittensor can be understood as the underlying infrastructure for projects like Covenant AI, providing incentives, governance, and network rules, rather than directly developing specific AI models or applications. Subnets like Covenant AI, on the other hand, act more as "application-layer builders" offering specific AI tasks and model capabilities on the underlying network.
Covenant AI's Sudden Announcement
On the morning of April 10, Sam Dare, the founder of Covenant AI, suddenly issued a statement (considering TAO's continuous decline, the conflict may have been brewing for longer), stating that due to Bittensor and its representative Jacob Steeves (online alias Const)违背去中心化理念 (violating decentralized principles), Covenant AI had decided to withdraw from the Bittensor network.
Covenant AI stated in its announcement that its core belief is that "the training of cutting-edge AI models should not be controlled by any single entity," but when a single actor can suspend subnet emissions, override a subnet owner's management rights over their own community space, publicly abandon projects without due process, and use token dumping as a coercive mechanism to force compliance, this is not decentralization but centralized control disguised as decentralization.
Covenant AI further alleged that every participant in the Bittensor ecosystem — miners, validators, and investors — should be aware that this power exists and has been exercised by Const. Const exercised this power not for the health of the network but to regain control over a team that had become "too independent" and difficult to manage — a subnet owner capable of building its own community, making independent decisions, and operating permissionlessly, as this threatened his power over the entire ecosystem. Specifically, while Bittensor adopts a so-called "triumvirate" structure, where three individuals manage network upgrades via multi-signature, and claims this is distributed governance to the community, the reality is different. Const实际上仍掌控绝对权力,且抵制任何真正的权力移交 (Const实际上仍掌控绝对权力,且抵制任何真正的权力移交) — the power in the Bittensor ecosystem has never left one person's hands.
Covenant AI also mentioned that over the past few weeks, Const had taken a series of actions against the team's operations that conflicted with the principles proclaimed by Bittensor, including suspending Covenant AI's subnet emissions, removing the team's administrative permissions for its own community channels, unilaterally abandoning subnet infrastructure, and exerting economic pressure through large-scale public token dumping during operational conflicts.
Therefore, Covenant AI decided to exit the Bittensor network. The team concluded by stating that decentralized, permissionless AI training is not a feature unique to Bittensor but a technological capability the Covenant AI team hopes to continue advancing. Covenant AI's research, team, models, and vision will continue to move forward, with very exciting projects currently underway, details of which will be announced to the public soon.
Public Conflict, Bittensor Mired in Controversy
Due to the success of Covenant-72B (SubNet-3), and the fact that the Covenant AI team also operates two other key subnets — Basilica (SubNet-39, positioned as an AI model evaluation/reasoning-related subnet) and Grail (SubNet-81, positioned as a more complex task-driven AI subnet) — the team holds a pivotal position within the Bittensor ecosystem. It is perhaps precisely Covenant AI's growing influence in terms of community, resources, and voice that triggered the "power struggle"矛盾 with Const.
With the public airing of their conflict, the Bittensor ecosystem quickly descended into a whirlwind of controversy.
On the product level, with Covenant AI's departure, the community began to question the future development and value of the Bittensor network. As one of the teams with the strongest technical narrative and tangible results in the current Bittensor ecosystem, Covenant AI's exit means this capability chain is being directly removed. Bittensor's technical progress and ecosystem activity in AI model training will face uncertainty, and the market's judgment of its long-term value has consequently become more cautious.
In terms of reputational impact, Bittensor's decentralization narrative is facing its biggest challenge since inception. Covenant AI's accusations strike at the very core of Bittensor's narrative — the "decentralized AI network." For Bittensor, which relies on the decentralization narrative to attract developers and computing power participants, the impact of this governance dispute far exceeds short-term price fluctuations and is more likely to shake the confidence of ecosystem participants.
On the brand level, Covenant AI has used this controversy to conversely overshadow Bittensor in the community's perception. Prior to this announcement, the market's general impression of "Huang's praise" was that it was directed at Bittensor, with few realizing that Covenant AI was the true protagonist, and even fewer knowing of the team's existence. As the事件发酵 (event发酵), Covenant AI's visibility is放大 (amplifying), while Bittensor is becoming the side perceived as "bleeding" in the community's impression.
As of the time of writing, Bittensor's official social media has yet to comment. Const, on his personal account, gave a vague response: "This event will propel Bittensor towards its first truly 'headless' (likely意指不依赖单一团队 meaning not relying on a single team), truly commoditized subnets... Thank you Covenant AI for making Bittensor more decentralized."
Beneath Const's response, a large number of Bittensor community users (especially TAO holders) are urging Const to provide a more detailed response to the allegations raised by Covenant AI, but Const has not yet replied further.
Odaily Planet Daily will continue to follow this matter. Stay tuned.
