While the broader cryptocurrency market navigates a complex landscape in June 2026, Bittensor is pressing forward with its most ambitious infrastructure expansion to date — doubling its subnet capacity from 128 to 256 just weeks after losing one of its most prominent development teams. The decision to scale aggressively following the acrimonious exit of Covenant AI represents a high-stakes wager that decentralized AI infrastructure can outgrow its governance controversies.
By Tomas Novak | June 1, 2026
As the global cryptocurrency market enters June 2026, the sector remains in a state of high-velocity evolution. Bitcoin is currently trading at roughly 72,889 dollars, while Ethereum holds a position at approximately 1,983.17 dollars. Within this context, the niche of AI-driven protocols has become a focal point for institutional capital, yet it is currently overshadowed by internal friction within its most prominent representative: Bittensor.
The Agentic Protocol
Bittensor represents more than just a blockchain; it is an ambitious attempt to create a global, decentralized marketplace for machine intelligence. At its core, the protocol functions as a competitive arena where different “subnets” compete for TAO emissions by providing specific computational services, ranging from text generation and image synthesis to data scraping and protein folding. This “agentic” architecture is designed to incentivize the production of the best possible AI outputs through a process of peer-to-peer validation.
The thesis behind Bittensor is that by decoupling AI development from the silos of Big Tech, the world can access a more transparent and permissionless form of intelligence. Miners provide the “brains,” and validators act as the “judges,” ensuring that only the most accurate or useful models receive rewards. This decentralized AI thesis has attracted immense interest, with the network’s total subnet staking growing from a modest starting point just one year ago to a staggering 620 million dollars today. However, as the network scales, the tension between its decentralized ideals and the practicalities of technical leadership has begun to surface.
Neural Network Integration
Despite the current governance headwinds, Bittensor has achieved technical milestones that were once thought impossible for a decentralized network. The most notable of these was the development of Covenant-72B, a 72-billion-parameter large language model. This model was trained on an incredible 1.1 trillion tokens, proving that the subnet architecture could indeed support the massive data and compute requirements of modern neural networks. The significance of this achievement was even acknowledged by Nvidia CEO Jensen Huang, who publicly referenced Covenant-72B as a noteworthy technical milestone in the advancement of distributed machine learning.
The subnet architecture has proven flexible enough to host diverse projects. Some of the most successful integrations include:
- Chutes (SN64): A subnet focused on providing efficient API access for AI models, currently generating several million dollars annually from paid API calls, according to industry trackers.
- Targon (SN4): A project that has secured enterprise partnerships reportedly bringing in eight figures per year, demonstrating real-world utility beyond speculative trading.
These successes highlight the potential of the network to integrate complex neural networks into a cohesive, incentivized framework. Yet, the very team behind the Covenant-72B milestone—Covenant AI—is now at the center of the protocol’s greatest crisis.
Token Utility
The TAO token serves as the lifeblood of the Bittensor ecosystem, acting as the mechanism for resource allocation and governance. The tokenomics of TAO are modeled after Bitcoin, featuring a hard cap and a halving schedule that reduces daily issuance. In December 2025, the network underwent a significant halving, which saw daily issuance drop from 7,200 TAO to 3,600 TAO. This scarcity has been a primary driver of the token’s market valuation, which reached approximately 2.626 billion dollars by late May 2026.
However, critics often point to a “revenue gap” when evaluating the project’s sustainability. While the market cap sits in the billions, the network-wide external revenue is currently estimated between 3 million and 15 million dollars per year. This discrepancy suggests that much of the current value is derived from speculative interest and internal staking dynamics rather than external demand for AI services. Currently, about 48 percent of all TAO remains staked in the root network, while subnets account for roughly 19 percent of the staked supply. The subnet token market cap has recently surpassed 1.5 billion dollars, reflecting a growing ecosystem that is still searching for a sustainable balance between emissions and actual revenue.
Potential Bottlenecks
The governance crisis erupted in April 2026 when Covenant AI, one of the network’s most prestigious development teams, announced its departure. Sam Dare, the lead figure at Covenant AI, issued a scathing critique of the project’s direction. He accused Bittensor co-founder Jacob Steeves of running what he termed “decentralization theatre”—a system where centralized control is masked by a veneer of permissionless participation. Following this fallout, Covenant AI withdrew from three key subnets: Templar (SN3), Basilica (SN39), and Grail (SN81).
The impact on the market was immediate. Covenant AI sold approximately 37,000 TAO tokens, worth roughly 10 million dollars at the time. This sell-off contributed to a sharp decline in the price of TAO, which plummeted from roughly 340 dollars to near 250 dollars, representing a 25 percent decline. The Opentensor Foundation responded by introducing a “Locked Stake” governance model, which requires time locks on staked TAO before it can be withdrawn. This move was intended to stabilize the network and prevent sudden mass exits, but it has also drawn criticism for further centralizing the protocol’s rules in response to a crisis.
Final Verdict
The question remains: Can Bittensor survive this crisis of confidence? The bullish case rests on the project’s undeniable technical momentum and institutional backing. Nvidia reportedly holds a substantial position in TAO, with the majority of that supply staked, signaling a long-term commitment from the world’s leading AI hardware provider. Furthermore, Grayscale filed an S-1 for a spot TAO ETF with the NYSE in December 2025, suggesting that institutional access is on the horizon. The network is also in the process of expanding its subnet capacity from 128 to 256, providing more room for innovation.
Conversely, the bearish case focuses on the governance risks and the revenue-to-valuation gap. If Bittensor cannot transition from a speculative incentive machine to a revenue-generating utility, its high market cap may be unsustainable. The Emissions Refactor, which went live on May 13, 2026, was a step toward restructuring rewards for higher-performing subnets, but the loss of top-tier talent like Covenant AI is a heavy blow. Ultimately, Bittensor‘s survival depends on whether it can prove that its “decentralization” is more than just a marketing slogan, but a robust system capable of weathering internal dissent and external competition.
The cryptocurrency market remains highly volatile. This article is for informational purposes only and does not constitute financial advice.
doubling to 256 subnets right after covenant walks out is either bold or reckless. still trying to figure out which one
the covenant ai situation was messy but bittensor kept shipping. 256 subnets is a serious technical undertaking, curious if the network can handle the validator load
^ good question on validator load. more subnets with fewer quality teams filling them just dilutes everything
tao has been one of the few ai tokens with actual on-chain activity to back the narrative. the covenant drama was overblown imo