Key Takeaways
- • Grayscale increased Bittensor (TAO) exposure in its AI crypto portfolio from 31.35% to 43.06%, reducing NEAR Protocol allocation from 26.54% to 24.43% and cutting Filecoin and Graph positions, while Render saw a slight increase to 15.77%.
- • TAO experienced a 20% price decline following Covenant AI's abrupt exit on April 10, 2026, triggering a governance crisis within the Bittensor subnet ecosystem and sparking centralization concerns that sent the token from recent highs to a $240–$270 support range.
- • Bittensor's Covenant-72B successfully trained a 72.7-billion-parameter AI model across 70+ distributed nodes without a centralized data center, validating permissionless network coordination and demonstrating decentralized inference cost advantages via Subnet 64.
- • TAO's market capitalization stands at $6.2 billion with the network expanding to 128 subnets; Targon Compute subnet generated $105,000 in weekly revenue, translating to a $5.5 million annualized run rate despite an $82 million fully diluted valuation.
- • Grayscale filed an updated SEC registration statement for a proposed Grayscale Bittensor Trust that would hold TAO directly and potentially convert to an ETF listed on NYSE Arca, signaling institutional commitment despite near-term price volatility.
Summary
Grayscale has significantly reweighted its AI-focused cryptocurrency portfolio toward Bittensor (TAO) in its latest rebalancing, raising the token's allocation to 43.06% from 31.35%—a material increase that reflects institutional confidence in the network's growth trajectory. The rebalancing reduced exposure to competing AI assets, including NEAR Protocol (down to 24.43% from 26.54%), Filecoin (9.86% from 13.77%), and The Graph (4.15%), while Render received a modest increase to 15.77%. Notably, no new assets were added or removed from the portfolio, indicating a concentrated rotation within the existing holdings rather than a broader sector shift.
The timing of Grayscale's increased TAO exposure coincided with significant turbulence in the token's price action. TAO experienced a sharp 20% decline following Covenant AI's announcement of its exit from the Bittensor ecosystem on April 10, 2026, triggering an acute governance crisis within the subnet structure. The selloff raised fundamental questions about network centralization and governance resilience, pushing TAO into a $240–$270 support range with technical indicators reflecting bearish momentum (RSI at 42) despite not yet reaching oversold conditions. At the time of the rebalancing announcement, TAO was trading around $258–$321, down significantly from recent highs near $330.
The underlying fundamentals supporting Grayscale's conviction appear rooted in Bittensor's subnet economy expansion and demonstrated technical capabilities. The network has expanded to 128 independent AI-focused subnets, each functioning as an open-source startup within the framework. A critical milestone came with the successful training of Covenant-72B, a 72.7-billion-parameter AI model distributed across 70+ anonymous nodes without centralized data center infrastructure. This achievement employed SparseLoCo technology to reduce inter-node communication by over 146x, validating that permissionless networks can coordinate large-scale model training while maintaining distributed validation through Gauntlet's loss evaluation and OpenSkill ranking systems. More commercially relevant, decentralized inference operations via Subnet 64 are already undercutting centralized cloud providers on cost.
Ecosystem growth metrics underscore the network's expanding utility narrative. Targon Compute, a single Bittensor subnet, generated $105,000 in weekly revenue, extrapolating to approximately $5.5 million annualized despite possessing only an $82 million fully diluted valuation. TAO's broader market capitalization stands at $6.2 billion. Grayscale's commitment extends beyond portfolio allocation; the fund manager filed an updated SEC registration statement for a proposed Grayscale Bittensor Trust designed to directly hold TAO and track its market price, with conversion to an ETF and NYSE Arca listing contingent on regulatory approval. This institutional framework signals conviction that extends beyond near-term price volatility, though the governance concerns unveiled by Covenant AI's departure represent a material technical and structural risk that institutional investors must monitor closely.
The timing of Grayscale's increased TAO exposure coincided with significant turbulence in the token's price action. TAO experienced a sharp 20% decline following Covenant AI's announcement of its exit from the Bittensor ecosystem on April 10, 2026, triggering an acute governance crisis within the subnet structure. The selloff raised fundamental questions about network centralization and governance resilience, pushing TAO into a $240–$270 support range with technical indicators reflecting bearish momentum (RSI at 42) despite not yet reaching oversold conditions. At the time of the rebalancing announcement, TAO was trading around $258–$321, down significantly from recent highs near $330.
The underlying fundamentals supporting Grayscale's conviction appear rooted in Bittensor's subnet economy expansion and demonstrated technical capabilities. The network has expanded to 128 independent AI-focused subnets, each functioning as an open-source startup within the framework. A critical milestone came with the successful training of Covenant-72B, a 72.7-billion-parameter AI model distributed across 70+ anonymous nodes without centralized data center infrastructure. This achievement employed SparseLoCo technology to reduce inter-node communication by over 146x, validating that permissionless networks can coordinate large-scale model training while maintaining distributed validation through Gauntlet's loss evaluation and OpenSkill ranking systems. More commercially relevant, decentralized inference operations via Subnet 64 are already undercutting centralized cloud providers on cost.
Ecosystem growth metrics underscore the network's expanding utility narrative. Targon Compute, a single Bittensor subnet, generated $105,000 in weekly revenue, extrapolating to approximately $5.5 million annualized despite possessing only an $82 million fully diluted valuation. TAO's broader market capitalization stands at $6.2 billion. Grayscale's commitment extends beyond portfolio allocation; the fund manager filed an updated SEC registration statement for a proposed Grayscale Bittensor Trust designed to directly hold TAO and track its market price, with conversion to an ETF and NYSE Arca listing contingent on regulatory approval. This institutional framework signals conviction that extends beyond near-term price volatility, though the governance concerns unveiled by Covenant AI's departure represent a material technical and structural risk that institutional investors must monitor closely.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.