The MAD Podcast with Matt Turck Benedict Evans: OpenAI’s Moat Problem & the Future of Software
145 snips
Mar 19, 2026 
Benedict Evans, independent tech analyst known for platform and economics analysis, returns to dissect AI’s big questions. He argues foundation models lack network effects and face commoditization. He explores ChatGPT’s shallow usage, why better models do not fix UX, the rise of improvised and ephemeral software, and the financial strain of hyperscalers’ massive CapEx.
AI Snips
Chapters
Transcript
Episode notes
ChatGPT Usage Is Mile Wide Inch Deep
- ChatGPT has very wide reach but shallow engagement: many users can't think of productive uses and only ~5% pay.
- Evans calls this a "mile wide, inch deep" usage problem that forces OpenAI to trade mindshare for tangible assets or products.
Better Models Don't Solve UX For Critical Tasks
- Incremental model quality improvements often don't change product usability for tasks that require correctness.
- Evans explains that going from 90% to 91% correctness still forces users to verify outputs, so UX and tooling matter more than marginal model gains.
Build From User Experience Not From Research Releases
- Accept that product teams become strategy takers: new model capabilities arrive and teams must adapt product plans.
- Start from use cases and user experience rather than building forward from every new research breakthrough.