The podcast by and for AI Engineers! In 2025, over 10 million readers and listeners came to Latent Space to hear about news, papers and interviews in Software 3.0. We cover Foundation Models changing every domain in Code Generation, Multimodality, AI Agents, GPU Infra and more, directly from the founders, builders, and thinkers involved in pushing the cutting edge. Striving to give you both the definitive take on the Current Thing down to the first introduction to the tech you'll be using in the next 3 months! We break news and exclusive interviews from OpenAI, Anthropic, Gemini, Meta (Soumith Chintala), Sierra (Bret Taylor), tiny (George Hotz), Databricks/MosaicML (Jon Frankle), Modular (Chris Lattner), Answer.ai (Jeremy Howard), et al. Full show notes always on https://latent.space Sponsorship and business inquiries: business@latent.space www.latent.space (https://www.latent.space?utm_medium=podcast)
What this show covers
drug discoveryscaling lawsAI for scienceAI infrastructureAI safetyprotein-ligand bindingscientific agentsAI for biologyDNA-encoded librariesphysical AIroboticsworld-action modelsdeploymentautonomyAI biology
What listeners will find here
How Leash Bio turned DEL chaos into AI-ready chemistry data
DNA-encoded libraries, protein-ligand binding, and scientific AI are at the center of this conversation with Ian Quigley of Leash Bio. The episode digs into how Leash gets clean, reproducible data from a notoriously messy assay format, and why that matters for modeling, benchmarking, and agentic scientific reasoning.
Can a million-hour world model make robots useful for real work?
General-purpose physical AI is the focus here, as Jason Ma explains Dyna-2 and the push toward robots with hour-long autonomy. The episode mixes model architecture, hardware choices, deployment lessons, and a live demo of Taku to show how Dyna is trying to move beyond flashy robotics demos.
Why AlphaFold Was a Milestone, Not the End of Protein Folding
Protein folding is still far from solved, and this panel digs into why AlphaFold’s success doesn’t mean biology has yielded to scale. Pushmeet Kohli and Sal Candido explore better data, better inductive biases, and what it would take to model proteins, cells, and disease more predictively.
Can AI discover new materials only by touching the real world?
This episode explores Periodic Labs’ push toward “synthesis superintelligence,” where AI learns from physical experiments, simulations, and the full workflow of materials discovery. Liam Fedus and Ekin Dogus Cubuk argue that smarter models alone won’t crack science without noisy, real-world feedback.
Alex Zhang on GPU kernels, recursive LMs, and why future AI may look like a swarm
Alex Zhang’s episode spans GPU optimization, recursive language models, and the systems wrapped around modern AI. It’s a research-heavy conversation about harness design, agent swarms, and why better scaffolding may unlock far more capability than bigger models alone.
Episodes
More like Latent Space: The AI Engineer Podcast
The a16z Show
Andreessen Horowitz
Big Technology Podcast
Alex Kantrowitz
A Question of Science
The Francis Crick Institute
No Priors: Artificial Intelligence | Technology | Startups
Conviction
Industrial AI Podcast
Robert Weber / Peter Seeberg
Comptoir IA 🎙️🧠🤖
Nicolas Guyon
The Gargle
The Bugle
Google DeepMind: The Podcast
Hannah Fry
Dwarkesh Podcast
Dwarkesh Patel
Lex Fridman Podcast
Lex Fridman