- TAO Times ⚡️
- Posts
- TAO Times #43: Alpha Token Indexes
TAO Times #43: Alpha Token Indexes
Yuma & TrustedStake launches, Covenant 72B, & subnet updates

TAO Times #43
TLDR;
Highlight of the Week: Institutional on-ramps land: Yuma launches Subnet Composite and Large Cap funds; TrustedStake debuts Universe (40+ subnets) and Top-15 indices for passive alpha exposure.
Ecosystem & Research: Covenant72B posts a competitive 72B / 1.2T checkpoint; TaoRevenue ships a profitability dashboard; Score/Sire outlines perception-driven agents priced by prediction-market P&L.
Subnet Updates: New/live/updated: Babelbit (SN59) speech translation; ReadyAI–Ipsos data tagging; SN63 scoring & immunity tweaks; Swarm (SN124) services model; RESI (SN46) ReDeFi stack; Zeus (SN18) API + 50% buyback, and more.
Media & Community: DSV Revenue Search #34 (SN10) + SN62 AMA; Hash Rate features Hone (SN5); Ventura Labs hosts Seby/RESI; Sami Kassab adds David Lawee and @BaznoCap.
Development Updates: OpenDev readies SDK v10, launches Pylon (Commit-Reveal v4 to any language, hotkey protection), enables auto-staking, and shifts registration to market-priced with enhanced immunity.
Highlights of the Week
⚡️ Yuma announced the launch of Yuma Asset Management, a platform that introduces managed subnet-token funds accessible to accredited investors and institutions to help gain diversified exposure into Bittensor subnets.
The first two funds announced are:
The Yuma Subnet Composite Fund: Similar to the “NASDAQ Composite” for subnet tokens, this fund is designed for market-cap weighted exposure across all active subnets;
The Yuma Large Cap Subnet Fund: Similar to the “Dow Jones Industrial Average” for subnet tokens, this fund is designed for targeted exposure to the top subnets.
This move signals a maturing capital market structure around Bittensor, effectively lowering the technical barrier to subnet exposure and passively onboarding institutional liquidity into the alpha token economy.
⚡️ The team over at TrustedStake released their alpha token index platform, launching the Bittensor Universe Index, an S&P 500–style investment vehicle for subnet tokens.
The two inaugural products include:
Bittensor Universe Index — covering 40+ subnets across the ecosystem for broad, one-click exposure to decentralized AI infrastructure.
Top 15 Subnets Index — a concentrated basket mirroring an S&P 100–style approach, designed for investors seeking focused exposure to Bittensor’s top-performing subnets.
TrustedStake, founded by early Bittensor contributors and crypto infrastructure engineers, operates as a non-custodial staking and allocation platform built on the Substrate Proxy Pallet. The platform enables trustless delegation, automated subnet allocation, and onchain reward distribution via child-hotkey mechanisms.
⚡️ Covenant released a progress update on Covenant72B, its first large-scale decentralized training run for a 72-billion-parameter LLM coordinated permissionlessly on the Bittensor. The model is being trained on 1.2 trillion tokens using 20+ global compute participants, each running a minimum of 8 × B200-class GPUs. This marks one of the largest open, verifiable pre-training experiments conducted entirely across decentralized infrastructure.
The Covenant stack comprises three interconnected platforms:
Templar: decentralized pre-training system using Gauntlet (on-chain incentive/reward allocation) and SparseLoCo (communication-efficient optimizer).
Basilica: decentralized compute marketplace designed to support large-scale model training.
Grail: decentralized reinforcement-learning post-training network.
The first checkpoint, now live on Hugging Face, already outperforms prior decentralized training runs and approaches performance parity with centralized baselines like MBZUAI’s K2 model on multiple benchmarks (ARC-C/E, HellaSwag, MMLU).
Covenant’s framework demonstrates that large-scale, permissionless pre-training can achieve competitive results while maintaining oncchain transparency, participant auditability, and 6% network-wide communication overhead.
Subnet & Ecosystem Updates, Research, & Media ⬇️