In May 2026, decentralized AI is transitioning from the narrative-building phase to the infrastructure validation phase. Within this sector, Bittensor and its native token TAO remain central, impossible to overlook.
According to Gate market data, as of May 25, 2026, TAO is priced at $273.7, with a 24-hour decline of approximately 1.93%. Its market capitalization stands at about $2.626 billion, and overall market sentiment is neutral. Over the past 90 days, TAO has surged 63.01%, rebounding from a low of $165.1 to a high of $379.2. However, it’s still down roughly 36.94% from its peak of $538.9 a year ago. This trajectory reflects a highly polarized market perception: optimists view TAO as "the Bitcoin of AI," while skeptics question whether its massive market cap is backed by real, matching revenue.
Dual Narratives: Subnet Expansion and Governance Crisis
The first half of 2026 brought dramatic swings to the Bittensor ecosystem.
On one hand, the pace of ecosystem expansion accelerated. Bittensor’s number of subnets has grown to 129, gradually forming a concentration at the top and a layered structure. Industry trackers report that, as of late March 2026, the total TAO staked across Bittensor’s subnets soared from about $74,400 a year ago to over $620 million—a staggering 833,000% increase. The total market cap of subnet tokens has surpassed $1.5 billion.
On the other hand, cracks in governance erupted during the same period. On April 10, 2026, Covenant AI—one of Bittensor’s top development teams and the creators of the 72-billion-parameter language model Covenant-72B—suddenly announced its exit from the network. Founder Sam Dare publicly accused Bittensor co-founder Jacob Steeves of "centralized control," labeling the entire decentralized AI narrative as a "decentralization performance."
Following this announcement, TAO’s price dropped significantly in a short span. Covenant AI’s founder then sold off his TAO subnet tokens, triggering cascading liquidations and further price declines.
It’s important to clarify: Covenant AI’s exit and subsequent token sales are verifiable facts on public record, and the price movements are confirmed by market data—not speculative analysis.
The Gap Between Scale Expansion and Revenue Reality
Structural Divergence Within the Subnet Ecosystem
Bittensor’s expansion shows clear phases. From just one subnet at the start of 2023, it broke past 100 subnets in April 2025, and stabilized above 128 by early 2026. The protocol’s subnet cap is set at 256, leaving room for future growth. After the first halving in December 2025, daily TAO issuance dropped from 7,200 to 3,600, tightening supply significantly.
However, scaling up doesn’t guarantee a parallel improvement in quality. Of the roughly $2.36 billion TAO staked, over 30% has flowed into subnets, indicating capital migration from the mainnet to the application layer. Yet, staking distribution reveals clear structural divergence: about 48% of TAO remains staked in the root network, while subnets account for only 19%. This means most TAO holders still prefer the lower-risk, more stable returns of root network staking.
The Core Contradiction in the Revenue Model
Bittensor’s biggest structural challenge lies in the sustainability of its economic model.
With 129 active subnets, Bittensor miners and validators earn TAO emission rewards for contributing "intelligence" within each subnet. Confirmed external revenue across the network ranges from $3 million to $15 million per year. Specifically, Chutes (SN64) generates about $5.5 million annually from paid API calls, while receiving tens of millions in TAO emissions. Targon (SN4) brings in over $10 million per year and has ties with major enterprise partners.
Analysts note that most subnets still heavily depend on TAO emissions to incentivize participants, with a clear gap between subsidies and external revenue. For example, Chutes’ $5.5 million in external income is dwarfed by the tens of millions in TAO rewards it receives.
As the December 2025 halving continues to impact supply, if subnet external revenue doesn’t grow significantly within a reasonable timeframe, the gap between inflationary incentives and real demand may widen further. However, on-chain activity for inference and agent deployment is increasing, which some observers see as early signs of progress toward revenue independence.
Here’s a comparative overview of key Bittensor ecosystem metrics:
| Metric | Data | Description |
|---|---|---|
| Active Subnets | 129 | Protocol cap: 256 |
| Total Subnet Staking | Over $620 million | 833,000% growth in one year |
| Subnet Token Market Cap | Over $1.5 billion | Templar, Quasar, and others stand out |
| Chutes (SN64) Annual Revenue | ~$5.5 million | Paid API inference services |
| Targon (SN4) Annual Revenue | Over $10 million | Includes enterprise partnerships |
| Network-wide External Annual Revenue | $3M–$15M | Significant gap vs. market cap |
| TAO Circulating Market Cap | ~$2.626 billion | As of May 25, 2026 |
| Fully Diluted Valuation | ~$5.8 billion | Total supply: 21 million TAO |
| Daily Issuance | 3,600 TAO | Post-December 2025 halving |
| Root Network Staking Share | ~48% | Preference for low-risk allocations |
Market Sentiment: Institutional Endorsements and Community Divisions
Multiple Perspectives: Continued Institutional Investment
Since 2026, institutional narratives around Bittensor have intensified. Here are key events (fact-based):
Grayscale submitted an S-1 filing for a spot TAO ETF to the New York Stock Exchange in December 2025. Grayscale also launched the Grayscale Bittensor Trust, providing a compliant channel for institutional investors.
Additionally, NVIDIA CEO Jensen Huang publicly acknowledged the training achievements of the Covenant-72B model within the Bittensor ecosystem, calling it a noteworthy technical milestone. Social Capital founder Chamath Palihapitiya also mentioned this achievement in a public podcast.
Bearish Views: Governance Centralization and Revenue Desert
Bearish arguments are logically consistent as well. The Covenant AI incident exposed a core contradiction: Bittensor’s Yuma consensus mechanism is theoretically a decentralized game system, but in practice, the staking behind top validator nodes is highly concentrated among early investors, the foundation, and addresses linked to the founders. This means the founders are not only rule-makers but, to a large extent, the biggest referees.
Other analysts point out that subnet operators, after receiving TAO rewards, have no obligation to return the value of AI models, data, or services to the ecosystem. They may choose to privatize their outputs and leave Bittensor, preventing TAO holders from effectively capturing value.
Here’s a comparison of mainstream bullish and bearish viewpoints:
| Bullish View | Bearish View |
|---|---|
| Grayscale ETF application, high institutional interest | Huge gap between TAO market cap and actual network revenue |
| Halving mechanism tightens supply | Subnet income relies on inflationary subsidies, low external paid share |
| Rapid subnet expansion, emergence of independent token economies | Governance power is highly concentrated, founders exert strong control over validator nodes |
| Public recognition from NVIDIA CEO | Developer exit risk: core contributors can leave with their achievements |
Four Key Questions
Has decentralization reached market expectations?
A significant gap remains. The Covenant AI incident revealed that while computing power can be distributed, network control and capital are highly concentrated. This reality creates structural tension with Bittensor’s "decentralized AI" narrative. Public information shows Jacob Steeves proposed restarting community voting governance on Discord, indirectly confirming the need for improvements in the current governance structure. The effectiveness of these reforms remains to be seen.
Is the token economic model sustainable?
Revenue pressure is real. The gap between Chutes subnet’s $5.5 million external revenue and the tens of millions in TAO emissions is an on-chain fact. While the current ratio can be sustained by market narrative for now, long-term sustainability depends on the growth rate of external paid demand. There’s not enough data yet for a definitive judgment.
What drives institutional entry?
Institutional capital flows mainly reflect allocation needs for the broader AI sector. Grayscale’s ETF application makes TAO one of the more liquid institutional investment targets within the decentralized AI narrative. However, it’s important to distinguish: institutional allocation is based on "optimism about AI, with TAO as a leading option," not necessarily a full endorsement of Bittensor’s current economic model.
Is the competitive positioning against centralized AI valid?
Differentiation exists, but needs validation. Bittensor’s Chutes subnet claims its AI inference services are priced lower than traditional cloud providers. If this cost advantage persists and grows, Bittensor will have genuine differentiated competitiveness in the AI infrastructure market. However, current revenue data shows this advantage hasn’t yet translated into a large-scale paying user base.
Industry Impact: Structural Reference for Decentralized AI
Zooming out from Bittensor to the broader industry, several trends are worth watching.
The integration of decentralized AI and blockchain is accelerating. AI agents’ need for payment infrastructure is becoming a new growth driver for on-chain economies. In decentralized compute, various decentralized GPU protocols are rapidly expanding, offering alternatives to traditional centralized cloud services. The global AI market was valued at $350–$400 billion in 2025 and is projected to reach $1.5–$2 trillion by 2030.
Bittensor occupies a unique position in this landscape. It’s not simply a GPU rental marketplace; it incentivizes AI intelligence output itself—creating a decentralized machine intelligence market through its subnet architecture. This differentiated positioning makes Bittensor complementary, not directly competitive, with other decentralized compute protocols.
Here’s a comparative overview of Bittensor and other leading decentralized AI projects:
| Project | Core Positioning | Differentiation |
|---|---|---|
| Bittensor (TAO) | Decentralized AI model incentive coordination layer | Subnet architecture incentivizes intelligence output, not hardware rental |
| Render (RENDER) | Decentralized GPU rendering | Focuses on rendering, expanding to general AI compute |
| Akash (AKT) | Decentralized cloud compute marketplace | Reverse auction pricing, general compute resource market |
Conclusion
Bittensor’s current situation is fundamentally a classic question about value anchoring: When a protocol, driven by grand industry narratives, a carefully designed token economy, and strong institutional backing, achieves a market cap far exceeding its actual revenue, is the market pricing forward-looking value discovery, or is it narrative-driven preemptive valuation?
The 833,000% surge in subnet staking from $74,400 to $620 million is certainly eye-catching. Yet, the gap between Chutes subnet’s $5.5 million annual revenue and tens of millions in TAO emissions, as well as the network-wide $3–$15 million external revenue versus a roughly $2.6 billion market cap, cannot be ignored. Grayscale’s ETF application represents a long-term institutional bet, but Covenant AI’s exit statement reminds the market that governance risks under the decentralization narrative are far from negligible tail events.
The next breakthrough for decentralized AI may not come from simply adding more subnets, increasing staking, or ramping up institutional allocations. Instead, it will hinge on achieving structural progress in two fundamental areas: revenue generation and credible governance. For market participants, closely tracking quarterly changes in subnet external revenue, post-halving inflation absorption, and the execution of governance proposals will be essential for evaluating the evolution of this central question.
