AI Investor Panel
summary
Investors report decentralized training is now viable, citing a 72B-parameter model trained on a fully decentralized network and startups claiming 10x+ faster and cheaper training and inference. About 100 early-stage teams pursue agentic payments while News Research’s Hermes hits tens of millions of downloads. Solana is named as a low-cost infra stack with APIs like paid.sh. Tokenization of model revenue surfaces as a real thesis, with examples like Hyperliquid burns 99% and a bold target of 1T-parameter by Q1 2027.
AI-written summary · about
chapters (9)
- 0:00Welcome
- 1:08Investing theses in crypto and AI
- 2:52Agentic commerce and AI training
- 5:19Agents driving on-chain activity
- 8:09Why crypto AI matters
- 10:12Token models vs equity in AI
- 14:43Qualities of successful founders
- 19:39Predictions for crypto AI in 2026
- 22:06Closing
related talks
17mInfrastructure of AI
49mThe Role AI & Crypto Play in Consumer Facing Applications
12mHow To Run AI At Scale Without Going Broke
28mDePIN & Privacy
33mTrillion-Token Pretraining
13mAccelerate USA 2026: TAO.com's Yoav Cohen and Sunrise's Saeed Badreg
14mPowering the Future of Decentralized AI
16mHyperbolic