Episode Summary
Executive Summary: The episode centers on a BitTensor/Covenant AI controversy that triggered a sharp Tau selloff, alongside a broader case for BitTensor’s subnet economy as a global, permissionless marketplace for AI work. The hosts then pivot to an in-depth demo of an OpenClaw/Claude-based “council of advisors” workflow for startup decisions, plus off-duty discussions on Microsoft product placement, Star Wars/animation, laptops, books, and open-ear headphones.
Main Topics: BitTensor/Covenant AI controversy and Tau price shock (Priority: 5/5): A major dispute emerged after Covenant AI alleged that a BitTensor co-founder interfered with its subnets and that Sam Dare sold Tau and subnet tokens, prompting concern about governance, decentralization, and possible rug-pull behavior. BitTensor subnet governance and decentralized incentives (Priority: 5/5): The hosts argue that BitTensor’s value lies in permissionless global participation, where miners, validators, and subnet owners compete on price and performance, but governance must evolve to prevent bad actors from undermining trust. Video Subnet 85 (Vidayo) demo and market use case (Priority: 4/5): Gareth Howells explains a video-processing subnet that upscales, compresses, colorizes, subtitles, and optimizes video for archives, streaming, and low-bandwidth markets, using miners and another subnet for compute. OpenClaw / Claude council-of-advisors skill (Priority: 4/5): Ola Lehman demos a multi-step LLM workflow inspired by Andrej Karpathy: multiple personas generate answers, peer-review them anonymously, and a chairman model consolidates the final recommendation for decisions like equity grants. AI tooling, business operations, and startup infrastructure (Priority: 3/5): The episode repeatedly frames AI tools, LLMs, and permissionless systems as ways to improve founder decision-making, global talent access, storage/compression efficiency, and operating leverage for startups. Off-duty media, gadgets, and culture (Priority: 2/5): The hosts discuss Microsoft product placement in TV, the new Maul/Star Wars animated series, a new MacBook Pro and the budget Apple Neo laptop, influential business/design books, and Shokz-style open-ear headphones.
Key Arguments: BitTensor’s recent selloff was driven by governance fears and token sales, not necessarily the collapse of the project itself. The Covenant AI incident shows that BitTensor still has human trust and control issues that need stronger decentralization and staking/governance rules. BitTensor’s broader thesis remains intact: permissionless global labor and compute can lower costs and expand opportunity across borders. Video is a huge, growing market, and subnet-based optimization can make archives and streaming dramatically cheaper and better. An LLM “council” can improve founder decisions by separating ideation, criticism, and final judgment through anonymized multi-agent review. Open, permissionless participation lets talented people in places like Vietnam join AI work without visas, approvals, or traditional employment gatekeepers. Winner-takes-most subnet designs increase efficiency and quality, but may reduce lower-rung participation opportunities for smaller miners. Good startups should use real products and services that create economic value; the hosts reject meme coins and pure pump-and-dump behavior.
Data Points: BitTensor token (Tau) price before news: ~$335 - Price mentioned before the Covenant AI controversy broke BitTensor token (Tau) price after news: $271 - Price mentioned after the controversy and selloff BitTensor market cap: $2.93B - Host cites approximate market cap after decline BitTensor subnets: 128 - Total available subnet slots discussed repeatedly Covenant AI subnets: 3 - Subnet triad used for pre-training, compute, and post-training Covenant AI subnet IDs: 3, 39, 81 - Subnets involved in the Covenant AI model stack Model size: 72B parameters - Covenant AI’s decentralized model size Video processing subnet number: 85 - Vidayo / Video Subnet 85 Video processing market growth: $175M in 2025 to $1.1B in 2032 - Gareth cites projected market expansion Video share of internet traffic: 85% - Used to justify the size of the video optimization market File-size reduction target: ~60% smaller - Claimed compression improvement at same perceptual quality Raised capital for example startup: $3M - Used in the Claude council equity example Founder equity in example: 85% - Jason’s hypothetical startup cap table Investor equity in example: 15% - Jason’s hypothetical startup cap table Recommended equity range: 0.75% to 1% - Council output for VP of Engineering offer Council overlap range: 0.5% to 1.5% - Range where council members broadly agreed Suggested option pool: 10% to 15% - Recommendation to create pool before making offer Rhode Island top income tax rate: 3.75% to 6% - Referenced during tax discussion Cyprus tax rate: 12.5% - Ola’s business/residency setup Berlin tax rate referenced: ~50% - Comparative burden before moving business Apple Neo laptop price: $600 - Discussed as a lower-cost Apple device MacBook Pro spend: $3,500 - Jason says he bought a new high-end MacBook Pro Shokz open-ear headphones price examples: ~$54 - Wirecutter/Amazon-like reference during discussion
Pivotal Quotes: "BitTensor is not as decentralized as it claims." — Covenant AI statement (quoted by Jason): Core allegation in the governance dispute "I think Jason has got a very good understanding of the situation." — Gareth Howells: Gareth validates Jason’s reading of the Covenant AI/token controversy "The recommendation 0.75 to 1%. Offer 0.75. Negotiate up to 1% ceiling if pushed." — Ola Lehman’s council report: Final output from the LLM council on startup equity compensation
Implications: BitTensor is proving its economic promise, but governance trust will determine whether it scales beyond early adopters. Meanwhile, AI “councils” and specialized subnets point to a future where founders outsource more decision-making and production work to global, permissionless systems.
About This Week in Startups
Jason Calacanis covers startups, tech, markets, media, and all the hottest topics in business and technology. He also interviews the world’s greatest founders, operators, investors, and innovators.