RiskReversal Pod Why VCs Are Betting Billions on AI Labs That Have No Product
Feb 16, 2026
Michael Dempsey, a research-driven VC focused on AI, robotics, biotech and crypto. Evan Karvounis, early crypto trader turned DeFi investor and Hyperliquid backer. They debate the rush from AI research to market, why big tech must overinvest in AI, commercialization and on-device inference. Then they pivot to crypto: Hyperliquid’s product-first perp exchange, token valuation versus protocol utility, and Bitcoin vs gold narratives.
AI Snips
Chapters
Transcript
Episode notes
VCs Are Funding Research At Scale
- Venture capital is funding frontier research at scale, compressing time from lab to market.
- That shifts durable value accrual toward products and second-order effects, not just models.
Next Decade: Market-Cap Destruction
- The next decade may focus more on market-cap destruction than expansion.
- Technological turnover creates winners and losers across large incumbents and opens durable opportunities.
Local Models Change Margins Fast
- Frontier models often become runnable locally within ~12–18 months after release.
- Local inference can collapse inference costs and shift margin profiles toward applications.