Best AI podcasts in 2026

The AI news cycle has gone from quarterly breakthroughs to weekly. Keeping up by reading is brutal — every model launch comes with a 40-page paper and a thread war on X. Podcasts are the way most working developers, founders, and curious listeners actually stay current, because a long conversation with the people who built the thing tells you more than a press release ever will.

This list is what we'd actually recommend in 2026, broken out by who you are and what you want from your listening.

TL;DR

• For working engineers: Latent Space, Machine Learning Street Talk, Dwarkesh Podcast.
• For founders and operators: The Cognitive Revolution, Hard Fork, Pivot.
• For weekly news without the hype: Hard Fork, Decoder.
• For daily updates in 30 minutes or less: The AI Daily Brief.
• For long-form interviews with builders: Lex Fridman Podcast, Dwarkesh Podcast, The Ezra Klein Show.

Latent Space

• Best for: AI engineers who want to stay current on tooling, agents, and infra.
• Standout features: deep technical interviews with the people building production AI systems; coverage spans foundation models, code generation, multimodality, GPU economics, and AI agent frameworks.
• Considerations: assumes a baseline of engineering literacy. Newcomers will find some episodes go straight into the deep end.

Machine Learning Street Talk

• Best for: technically rigorous listeners who want the research-paper version of AI commentary.
• Standout features: Tim Scarfe and rotating co-hosts cover scaling laws, mechanistic interpretability, alignment research, and reasoning model internals; episodes regularly run two to three hours and treat the audience like graduate students.
• Considerations: not a casual listen. If you don't already know what a transformer is, start somewhere else and circle back.

Dwarkesh Podcast

• Best for: people who want to hear unfiltered conversations with the researchers and engineers actually building frontier models.
• Standout features: long-form, meticulously prepared interviews with figures from OpenAI, Anthropic, DeepMind, and the broader research community. Dwarkesh Patel is famous for reading every paper a guest has ever written before recording.
• Considerations: episode length is usually two to four hours. Treat them like documentaries rather than commute fillers.

The Cognitive Revolution

• Best for: founders, product leaders, and investors trying to figure out what AI does to their business.
• Standout features: host Nathan Labenz interviews builders, researchers, and operators across the AI stack — model labs, application companies, evaluation tooling. The mix of technical and economic framing makes this the rare show that works for both engineers and execs.
• Considerations: episode lengths vary wildly. Skim the show notes before committing to a 90-minute conversation that may or may not match what you came for.

Hard Fork

• Best for: a weekly podcast that treats AI as one of several big tech stories rather than the only thing happening.
• Standout features: New York Times tech journalists Kevin Roose and Casey Newton walk through the week's AI, social media, and tech news with a healthy mix of skepticism and curiosity. Strong on contextualising what a launch actually means versus how it's being marketed.
• Considerations: more news commentary than deep technical analysis. If you want benchmarks and architecture diagrams, this isn't the show.

Decoder with Nilay Patel

• Best for: industry strategy and the business models around AI.
• Standout features: long-form interviews with CEOs, regulators, and operators about how technology actually gets bought, sold, regulated, and shipped. The framing question is always some variant of "how does this business actually work?"
• Considerations: not exclusively about AI — covers the broader tech industry — but AI threads through almost every recent episode.

Pivot

• Best for: bigger-picture takes on tech, media, and business with an AI undercurrent.
• Standout features: Kara Swisher and Scott Galloway run a fast, opinionated, twice-weekly format. AI shows up in almost every episode, usually in the context of capital markets, leadership, or industry consolidation.
• Considerations: if you want pure AI content without business commentary, you'll find Pivot's signal-to-AI-noise ratio lower than the more focused options on this list.

The AI Daily Brief

• Best for: a 20–30 minute drumbeat of what happened in AI yesterday.
• Standout features: hosted by Nathaniel Whittemore (often credited as Craig Fuller's replacement on similar formats), each episode summarises the day's most important AI stories with quick context. Great for keeping up without committing hours.
• Considerations: news-velocity format means episodes go stale fast. Listen the day they drop or skip.

Lex Fridman Podcast

• Best for: marathon interviews with the people behind some of AI's biggest moments.
• Standout features: Lex Fridman has hosted founders and researchers from across the AI industry, plus a wider cast of scientists, philosophers, and engineers. The format works well for understanding what someone is actually thinking, since the conversations are unhurried.
• Considerations: episodes run three to five hours. Skim chapter markers and jump to the parts you care about — listening front-to-back rarely makes sense.

The Ezra Klein Show

• Best for: thoughtful, policy-shaped discussions of AI's social and economic implications.
• Standout features: Ezra Klein brings a journalist's instincts to long-form interviews with researchers, ethicists, and policy thinkers. Episodes about AI tend to focus on what new capabilities mean for work, education, and democratic institutions.
• Considerations: less interested in technical detail than in second-order effects. Pair with Latent Space or Dwarkesh for a complete picture.

How we chose

We weighted four things: how much working AI practitioners actually listen to the show, how recent the content is (AI podcasts age fast), how much new technical or strategic information each episode contains, and host pedigree. We pulled candidate shows from the iTunes Charts technology category, our own Podtastic data on which AI shows our listeners follow, and direct community recommendations from working ML engineers.

If you're new to the space, we'd start with Hard Fork for context, The Cognitive Revolution for builder interviews, and Latent Space when you're ready to go technical. Those three together cover most of the surface area without overwhelming you with hours of listening per week. Once you've got those running, our tips for catching up on a podcast backlog help when subscriptions start outpacing your listening time.

For broader tech recommendations beyond AI specifically, our best tech podcasts list covers the wider shows in the category.

Listen smarter with Podtastic

Get more out of your new podcast picks. Podtastic is a fully featured podcast player for iOS and Android, built around Pod-telligence — a set of AI features that helps you get more out of every show:

• Smart Summaries — AI summaries of every podcast and episode so you know what's coming before you hit play
• Smart Topics — key topics surfaced across your favourite shows so you can jump straight to what matters
• Smart Playback — your queue fills itself based on what you actually listen to
• Jump Ahead — automatically tightens gaps and pacing so episodes flow naturally

Join the waitlist at podtastic.app to get early access.