Bittensor Drama! TAO down 15%! | E2274
This episode explores the recent Bittensor governance crisis where a prominent subnet founder allegedly mishandled tokens, causing TAO to drop 15%, and examines why decentralized AI networks offer builders a permissionless way to tap global talent and compute. Jason Calacanis and guest Gareth Howles discuss how Video Subnet 85 uses distributed mining to upscale and optimize video—illustrating the real product potential beneath crypto's technical complexity—while also demonstrating an AI council of advisers tool that uses multiple personas to pressure-test founder decisions.
Key takeaways
- • Decentralized subnets remove geographic friction for talent: developers in Vietnam, Thailand, and Africa can earn competitive rates directly without needing visas, employment contracts, or payment processing delays.
- • Buying TAO is like buying an ETF into distributed AI startups—you gain exposure to 128+ different subnets solving real problems (video processing, compute, training) without picking winners; this is a lower-risk entry point than betting on individual subnet tokens.
- • Structure equity offers with documented option pools first: create a 10–15% option pool and get board approval before negotiating offers (e.g., 0.75–1% for a VP of Engineering), or the cap table will shift mid-conversation and kill credibility.
- • AI can simulate a board of advisers: a single model like Claude Opus can spawn personas (contrarian, expansionist, first-principles thinker, executor, outsider) that debate decisions in parallel, get anonymized peer review, and synthesize a final recommendation—saving founders cognitive load and blind spots.
- • Video upscaling and compression solve real customer problems at scale: 85% of internet traffic is video; old archives (BBC, Getty, CBS) need 4K upscaling, colorization, and compression for streaming—a market projected to grow from $175M to $1.1B by 2032.
- • Avoid clumsy product placement: Microsoft's heavy-handed Copilot integration in the TV show *High Potential* damages credibility and pulls viewers out of the narrative—building great products and letting them speak for themselves is a better long-term strategy than aggressive advertising.
Recommendations (13)
"it has some of the best world building and characterization I've ever read. I'm absolutely hooked. It's the type of book where I don't want to go to bed because I want more of it."
This Week in Startups · ▶ 1:16:59
"I can recommend to you the Shokz open ear headphones when you're walking with the baby, you need to hear if the baby wakes up"
This Week in Startups · ▶ 1:17:43
"Nigel Holmes is the godfather of design. You could pull up this book, Designer's Guide to Creating Charts and Diagrams."
This Week in Startups · ▶ 1:13:55
"I'm trying to Ogilvy on advertising. This was like an old book on advertising."
This Week in Startups · ▶ 1:14:52
"My life in advertising and scientific advertising by these are two works by Claude Hopkins."
This Week in Startups · ▶ 1:15:00
Mentioned (20)
More from these creators
Anthropic’s Mythos is a cyber-weapon, so you can’t have it | E2273
3 AI Agents That Actually Replaced Human Jobs | E2272
AI Rebuilt Every YC W26 Startup. Should Founders Be Scared? | E2271
Venture Roundtable: SpaceX IPO, AI's PR Crisis, and the Defense Tech Bubble | E2270
The 5-Step Framework for AI Agents That Improve While You Sleep | E2269
The $60 billion resource hiding in space, and the start trying to mine it | E2268