Bittensor governance conflict triggers TAO to plummet 25%: Decentralized AI narrative faces scrutiny

On April 9, 2026, the decentralized AI network Bittensor faced its most severe governance crisis since its inception. Its core subnet developer Covenant AI announced its exit from the ecosystem and publicly accused co-founder Jacob Steeves of holding “centralized control” over the network. Following the news, Bittensor’s native token TAO plummeted from $337 to around $250 within six hours, a 25% drop, evaporating approximately $650 million in market value and triggering about $9.1 million in forced long liquidations. As of April 15, 2026, TAO’s price remained around $250, with market confidence significantly shaken.

Why the core developers are accusing Bittensor’s decentralization of being a performance

Covenant AI founder Sam Dare openly stated that Bittensor’s so-called “multi-signature governance” is essentially a “decentralized theater.” He accused Jacob Steeves of exerting effective control over the three-person ruling team, capable of bypassing consensus unilaterally to implement network changes at any time, with other members merely serving as legal shields.

Specific allegations include: Steeves paused token emissions for Covenant AI’s subnet, deprived its community channels of management rights, publicly deprecated subnet infrastructure without following established procedures, and exerted economic pressure through large-scale token sales during conflicts. Dare sharply remarked in his statement: “The entire core promise of Bittensor—that no single entity can control it—is a lie.”

Notably, Covenant AI is not a marginal player in the ecosystem. The team previously completed Covenant-72B—a pretraining project of a decentralized large language model with 72 billion parameters, collaboratively developed by over 70 independent contributors on general hardware, recognized publicly by Nvidia’s CEO and cited by Anthropic’s co-founder. Such a leading developer’s public departure constitutes a substantial blow to Bittensor’s credibility.

How co-founders respond to centralization accusations and attempt to rebuild trust

In response to the allegations, Jacob Steeves did not deny the existence of the “three-person governance” structure. He acknowledged that Bittensor’s governance documents indeed describe a transitional governance model where a “three-person ruling council” composed of Opentensor Foundation staff and the Senate jointly hold root permissions.

Steeves pointed out that when dTAO was launched a year ago, the team initially planned to implement a community-led subnet governance mechanism—subnet alpha holders could vote via wallets to decide hyperparameters. However, this feature was delayed because early on it required giving subnet owners more control. He stated on Discord that now is the time to restart this discussion, suggesting the introduction of a community voting mechanism to elect the team and reboot the subnet.

Meanwhile, Steeves harshly criticized Covenant AI’s actions, calling Dare’s decisions “obviously malicious and greedy,” and claiming that his departure was aimed at causing “maximum pain.” He shifted the focus of the controversy from governance structure itself to accountability for individual actions.

Market mechanisms behind the 25% TAO price drop within six hours

The price plunge was not solely caused by governance disputes. Market data shows TAO fell from $337 to $253 within six hours, wiping out over $650 million in market value. However, deeper market impacts stem from chain reactions.

About six hours after Dare’s exit announcement, he sold approximately 37,000 TAO from his wallet—worth about $10 million. Analyst Michaël van de Poppe pointed out that the real destructive force was not the governance allegations themselves but the panic selling and leverage liquidations triggered by this sale.

This selling pressure then triggered a negative spiral: validator staking declined, reducing consensus weight; decreased rewards weakened incentives; and eroded market confidence accelerated user outflows. Trading volume surged to $1.72 billion on April 10, a significant increase from the roughly $500 million average earlier in the month.

Can the staking lock-up proposal prevent similar conflicts at the mechanism level?

In response to the controversy, Steeves introduced a new protocol feature called “locked staking,” regarded by some observers as one of the most important proposals in Bittensor’s history.

The core idea is to shift governance from interpersonal trust to cryptographic commitments enforced by on-chain code. Subnet owners are required to lock tokens for a fixed period, during which tokens cannot be transferred, providing the community with a verifiable long-term commitment signal. The principle of “time plus staking equals trust” applies—the product of staked amount and remaining lock time determines the strength of the commitment; larger holdings and longer locks imply higher ownership weight.

Steeves admitted that his real mistake was not implementing this supplementary mechanism earlier, and he believes that if the lock-up staking had been introduced sooner, it might have prevented the parties from splitting. If approved, this proposal would provide an on-chain, auditable guarantee of subnet owners’ long-term commitments, preventing sudden exits like Covenant AI’s and large-scale sell-offs from recurring.

However, lock-up staking alone cannot resolve the governance power distribution issues inherent in the “three-person ruling” structure. It is more of an economic incentive patch than a fundamental reorganization of power.

What structural challenges does Bittensor’s valuation logic face?

The governance dispute has brought Bittensor’s valuation foundation into focus. As of March 2026, its circulating market cap was approximately $2.6 to $3.6 billion, with a fully diluted valuation around $5.8 to $7 billion. Yet, the network’s actual external revenue—cash flow from user payments for AI models and services—is far below market valuation.

For example, the largest subnet, Subnet 3 (Templar), receives about $52 million in TAO inflation subsidies annually, but its external real revenue is only about $2.4 million—more than 20 times less. This indicates that the protocol’s current valuation heavily relies on inflation-driven internal economic cycles rather than genuine external commercial value capture.

The outbreak of governance conflicts further exposes this structural fragility: when a network’s trust basis is built on market consensus around the “decentralization” narrative, and that narrative is publicly questioned by core builders, the valuation premium faces the risk of re-pricing.

How should governance power be allocated in decentralized AI networks?

The core question raised by the Bittensor incident is: how should governance power be distributed in an AI network that claims to be “permissionless” and “decentralized”?

Covenant AI’s accusations reveal a deeper contradiction: transitional governance structures can improve decision-making efficiency early on, but over the long term, the tension between centralized control and decentralization intensifies as the ecosystem grows. Steeves admits the “three-person governance” exists but argues it is a transitional arrangement toward fully open governance.

The controversy centers not on whether a transition period should exist but on: who decides its duration, who controls the pace of decentralization, and whether fair arbitration mechanisms exist when builders and core teams disagree. These issues are not only about Bittensor’s future but also reflect common governance challenges faced by the entire decentralized AI sector.

Summary

The core value of Bittensor’s governance conflict lies not in the price fluctuations themselves but in exposing the gap between the decentralization narrative and actual practice. Covenant AI’s accusations brought the “three-person governance” structure into public view, while the 25% drop in TAO reflects market sensitivity to the credibility of that narrative.

Mechanically, the locked staking proposal represents an evolution from legal accountability toward cryptographic enforcement, but it cannot fundamentally resolve power distribution issues. From a valuation perspective, Bittensor remains subsidy-driven, with a significant gap between external real revenue and market cap. The governance dispute may accelerate market reassessment of this structural contradiction.

For the decentralized AI sector, the Bittensor incident offers a key lesson: decentralization is not a one-time declaration but an ongoing institutional practice. The network’s governance architecture must align with its narrative commitments; otherwise, even the best builders will vote with their feet.

FAQ

Q1: Why did Bittensor’s TAO tokens sharply decline on April 10?

The decline was caused by multiple factors: Covenant AI announced its exit and accused Bittensor governance of centralization, triggering market panic; its founder then sold about 37,000 TAO (worth around $10 million), intensifying selling pressure and causing approximately $9.1 million in forced long liquidations. TAO fell from $337 to $253 within six hours, a 25% drop.

Q2: What is the “three-person governance” structure?

Bittensor’s governance documents describe a transitional model where a “three-person ruling council” composed of Opentensor Foundation staff and the Senate jointly hold root permissions. Covenant AI accused this structure of being effectively controlled by Jacob Steeves, who can bypass consensus to implement changes unilaterally.

Q3: How does the locked staking mechanism work?

Locked staking is a new protocol feature proposed by Bittensor’s co-founders in response to this crisis. Subnet owners must lock TAO tokens for a fixed period, during which tokens cannot be transferred. The product of staked amount and remaining lock time measures the strength of the long-term commitment. This mechanism aims to replace personal trust with on-chain code enforcement to prevent large-scale sell-offs or sudden exits.

Q4: What impact does this incident have on the decentralized AI sector?

The incident exposes common governance challenges: how to balance early decision-making efficiency with long-term decentralization, and how to ensure “decentralization” commitments align with actual governance practices. The success or failure of the locked staking proposal could provide important mechanism references, but fundamental governance power distribution issues remain for the industry to explore.