Bittensor TAO Rallies 90% on Covenant-72B Milestone as Grayscale Files Spot ETF Amid Valuation Concerns

Bittensor's TAO token has surged 90% to trade at $334 following the successful training of Covenant-72B, a 72.7-billion-parameter model built on a decentralized network of 70+ anonymous nodes according to reports. This technical milestone demonstrates that distributed systems can coordinate large-scale model training without centralized data centers, challenging the necessity of co-located hardware for frontier AI development as documented. Concurrently, Grayscale Investments has increased its allocation to 43% and filed for a spot ETF, signaling strong institutional interest despite ongoing economic risks within the network according to market analysis.

The network achieved a 67.1 MMLU zero-shot score, outperforming established models like LLaMA-2-70B and LLM360 K2 through the use of SparseLoCo, which reduced inter-node communication by over 146x. While the technical achievement is significant, the immediate commercial narrative focuses on inference costs, where Subnet 64 is already undercutting centralized cloud providers according to technical benchmarks. This shift from pretraining limitations to commercial cost advantages has re-anchored the token's price structure around the 200-day moving average at $275 as market data shows.

Despite the bullish momentum, the asset experienced a sharp 27% correction to $263 earlier this month after Covenant AI founder Sam Dare exited the network, wiping out $900 million in market value according to financial reports. This event underscores the volatility inherent in decentralized AI ecosystems and their dependency on key figurehead participation as data indicates. The recent 140% surge in 24-hour trading volume and record futures open interest of $554.98 million highlight the intense speculation surrounding the ETF filing according to market analysis.

How Does Covenant-72B Validate Decentralized AI Training?

The completion of Covenant-72B represents a pivotal moment for the Bittensor ecosystem by proving that anonymous contributors can produce models competitive with billion-dollar corporate labs according to industry analysis. The model utilized sparse communication techniques and 2-bit quantization to overcome the traditional barriers of VRAM synchronization physics that usually require co-located hardware as technical reports state. Although frontier pretraining still faces physical constraints, the success of this 72-billion-parameter model validates the architecture for future distributed training efforts according to experts.

The technical implementation of SparseLoCo allowed the network to coordinate 70+ nodes effectively, reducing communication overhead by a factor of 146 compared to traditional methods according to technical documentation. This efficiency gain is critical for scaling decentralized networks, as it allows for the training of massive models without the prohibitive costs of centralized infrastructure as data shows. The resulting model performance, with a 67.1 MMLU score, places it ahead of major open-source benchmarks, suggesting a viable path for decentralized AI production according to performance metrics.

What Are The Economic Risks And Institutional Drivers?

Grayscale's decision to raise its Bittensor allocation to 43.06% and file for an NYSE Arca spot ETF aims to lower barriers for traditional capital inflows according to institutional reports. This move is supported by other institutional players like Polychain Capital, which has allocated $200 million, and pending filings from Bitwise according to market data. The ETF filing has already triggered a massive increase in market activity, with derivatives data showing positive funding rates and record open interest as observed.

However, significant valuation risks persist due to a severe subsidy imbalance where network subsidies reach $52 million against only $2.4 million in annual external revenue from the top subnet according to financial analysis. Token emissions continue to dilute holders despite the 2025 halving, complicating the assessment of genuine network growth versus emission-inflated metrics as data indicates. Investors must monitor whether external demand can grow faster than token emissions to sustain long-term value given these structural economic pressures according to market analysis.

How Does Leadership Volatility Impact Market Stability?

The market remains sensitive to leadership changes, as evidenced by the 27% crash to $263 triggered by the exit of Covenant AI founder Sam Dare according to market reports. This event resulted in approximately $900 million being wiped from the market capitalization, highlighting the volatility of assets tied to specific figureheads as financial data shows. Such movements demonstrate that decentralized networks, despite their distributed nature, still exhibit high correlation with the actions of key community leaders according to analysis.

While the technical foundations of the network have proven robust with the Covenant-72B launch, the price action reflects a tension between technical progress and market sentiment driven by human factors as market data indicates. The recent correction to the $263 level was quickly followed by a recovery as the ETF narrative gained traction, suggesting that institutional drivers may eventually outweigh individual leadership volatility according to market analysis. Long-term stability will depend on expanding the staking ecosystem and ensuring network security independent of any single entity according to experts.

Market analysts note that a close above $275 serves as a macro floor, while a break above $379 with volume could target the $420–$450 range contingent on Bitcoin stability according to technical analysis. The 12% correction seen in early April 2026 brought the network back to the $300 mark, providing a stabilization point for the subsequent rally according to market data. As the network navigates these economic and structural challenges, the focus remains on whether decentralized inference can deliver sustained commercial utility to offset token dilution according to industry reports.