The a16z Show The 2045 Superintelligence Timeline: Epoch AI’s Data-Driven Forecast
516 snips
Nov 24, 2025 
David Owen and Yafa Edelman from Epoch AI dive into the rapid evolution of artificial intelligence. They discuss why Anthropic may lead in data-center capabilities and the potential for AI to tackle complex mathematical problems like the Riemann hypothesis in mere years. They caution about job market shifts, exploring the disappearance of 10% of jobs this decade. The duo assesses the balance between AI-driven growth and the risks of a market bubble, emphasizing the economic implications of AI on productivity and GDP.
AI Snips
Chapters
Transcript
Episode notes
Spending Signals Real Economic Value
- Companies are spending massively on compute because inference is already profitable and users pay for value.
- This spending suggests current growth is real, not merely a speculative bubble.
Profit Today, Investment For Tomorrow
- Current profit trends show labs could be profitable if they paused new development.
- Ongoing heavy R&D investment aims to secure future growth rather than cover past losses.
Why Software-Only Takeoff Is Uncertain
- A software-only singularity requires AI to automate R&D faster than experimental compute scales.
- Current evidence shows experimental compute spending remains very large, arguing against pure software takeoff.