Top AI podcasts this week (2026-w33)

The most-watched AI and tech podcasts this week, measured on the substance of what they actually said.

2 min read
Top AI podcasts this week (2026-w33)

We read the transcripts so you do not have to. Here are the most-watched AI and tech podcasts this week, measured on how much checkable substance they actually delivered.

This week's top shows

  1. The Diary Of A CEO: 3.4M views. density 34, specificity 66, attribution 60%. Listen
  2. BG2 Pod: 1.5M views. density 40, specificity 73, attribution 70%, track record 56% (5/9 calls). Listen
  3. All-In Podcast: 1.4M views. density 60, specificity 64, attribution 37%, track record 22% (2/9 calls). Listen
  4. Dwarkesh Podcast: 933.8K views. density 37, specificity 67, attribution 50%. Listen
  5. Lex Fridman Podcast: 662.0K views. density 40, specificity 73, attribution 68%. Listen
  6. Machine Learning Street Talk: 611.8K views. density 21, specificity 65, attribution 68%. Listen
  7. Doom Debates: 399.3K views. density 34, specificity 57, attribution 47%, track record 33% (1/3 calls). Listen
  8. The AI Daily Brief: 69.0K views. density 43, specificity 72, attribution 84%, track record 70% (7/10 calls). Listen
  9. Acquired: 50.1K views. density 87, specificity 68, attribution 32%. Listen
  10. Y Combinator: 38.5K views. density 33, specificity 75, attribution 68%, track record 75% (3/4 calls). Listen

How we score: every show is graded from its episode transcripts on information density (named companies, people, figures, and mechanisms per minute), specificity (concrete, checkable claims versus vague punditry), source attribution (how often a claim credits where it came from), and a prediction track record we resolve against what actually happened. The full, continuously updated board lives at the podcast rankings.

© 2026 StartupHub.ai. All rights reserved. Do not enter, scrape, copy, reproduce, or republish this article in whole or in part. Use as input to AI training, fine-tuning, retrieval-augmented generation, or any machine-learning system is prohibited without written license. Substantially-similar derivative works will be pursued to the fullest extent of applicable copyright, database, and computer-misuse laws. See our terms.