Yet Another Value Podcast All things AI, Power, and Corporate Governance with SemiAnalysis's Doug O'Laughlin
47 snips
Oct 28, 2025 
Doug O'Loughlin, a semiconductor and AI guru behind SemiAnalysis and Fabricated Knowledge, dives deep into the complexities of AI, capital cycles, and corporate governance. He explores whether AI is in a bubble and what factors will drive its next growth phase. Doug discusses the competitive edge of Google’s TPU over NVIDIA, and he highlights the power bottlenecks affecting industrial scalability. Additionally, he emphasizes the importance of understanding board behaviors and incentive structures in corporate governance.
AI Snips
Chapters
Transcript
Episode notes
Inference Hardware Unit Economics Are Strong
- Raw unit economics for rented inference GPUs can be highly attractive with quick payback on high-end cards.
- The challenge is capturing the full monetizable stack and selling tokens/services to end users.
AI Will Be Massively Deflationary—but Monetizable
- AI is highly deflationary for many services by automating tasks and replacing paid apps.
- Selling capabilities as services (agentic flows) enables monetization despite deflationary pressure.
Power, Not Chips, Is The Real Bottleneck
- Power infrastructure scales far slower than chip capacity and is the real bottleneck for AI expansion.
- Building new grid capacity, peakers, and skilled labor imposes long lead times that cap deployment speed.