Video summary
State of AI in 2026: LLMs, Coding, Scaling Laws, China, Agents, GPUs, AGI
In this Lex Fridman conversation with Sebastian Raschka and Nathan Lambert, the discussion centers on the state of AI heading into 2026. The excerpt covers major model releases, the impact of the DeepSeek moment, open-weight models, competition between U.S. and Chinese labs, and how products like Claude, Gemini, and ChatGPT are shaping user behavior. It also touches on why organizational culture, hardware budgets, and real-world usage patterns may matter as much as raw model quality.
State of AI overview
A wide-ranging discussion on the current state of AI, with a focus on recent breakthroughs and what may come next.
U.S. vs. China AI race
The conversation examines competition between U.S. and Chinese AI labs, including open-weight model strategies and market influence.
Models, coding, and usage
The guests compare major model releases, coding performance, user adoption, and how brand, memory, and workflow shape usage.
Topics
State of AI in 2026
A broad look at the current AI landscape, including recent technical progress and what may happen next year.
China and open models
The guests discuss DeepSeek, open-weight models, and the growing number of strong Chinese AI labs.
Model competition and usage
The conversation compares Claude, Gemini, and ChatGPT, including hype, adoption, and coding-focused workflows.
Start with the video endpoint to capture ID, channel, publish date, duration, and source context.
Pull timestamped transcript data for summarization, search, citation, and RAG preparation.
Collect visible audience comments to identify themes, objections, questions, and engagement signals.
Persist structured JSON, run analysis, and publish dashboards, alerts, or research reports.
Public transcript excerpt
Transcript
Timestamped public transcript passages group captions into readable sections, making the video easier to scan, cite, and summarize.
Show timestamped transcript excerpt(1 passage)
on. And now, dear friends, here's Sebastian Raschka and Nathan Lambert. So I think one useful lens to look at all this through is the so-called DeepSeek moment. This happened about a year ago in January 2025, when the open-weight Chinese company DeepSeek released DeepSeek R1, that I think it's fair to say surprised everyone with near-state-of-the-art performance, with allegedly much less compute for much cheaper. And from then to today, the AI competition has gotten insane,
Related Crawlora APIs & guides
Build YouTube data workflows with Crawlora
This showcase is built from Crawlora's public YouTube data APIs. Use the same endpoints and guides to build your own transcript, comment, and creator-intelligence workflows.
More AI video examples
Browse structured transcript and comment showcases in AI.
More Podcasts video examples
Browse structured transcript and comment showcases in Podcasts.
YouTube API
Transcript, comments, and video metadata endpoints that return normalized JSON.
YouTube transcript extraction
Build searchable, RAG-ready transcript pipelines from public videos.
YouTube creator intelligence
Monitor creators, audiences, and content trends across channels.
Podcast & audio intelligence
Turn long-form audio and podcasts into structured, analyzable data.
Related showcases
More structured YouTube examples
Jensen Huang on NVIDIA, Extreme Co-Design, CUDA, and the AI Revolution
Jensen Huang explains NVIDIA’s move from GPU acceleration to full-stack AI infrastructure, focusing on extreme co-design, distributed computing challenges, and the strategic evolution that led to CUDA and a broader computing platform.
Jeff Kaplan on Warcraft, Overwatch, Blizzard, and the Making of Online Worlds
Jeff Kaplan reflects on the arcade, console, and PC games that shaped his love of gaming, including Pac-Man, Zork, Quake, and EverQuest. The excerpt follows his path from player to Blizzard designer, his emotional departure from the studio, and a preview of his new open-world multiplayer game set in 1800s California.
Rick Beato on Hendrix, Django Reinhardt, Bebop, and Ear Training
Rick Beato discusses early guitar inspiration, Hendrix, Django Reinhardt, bebop, and how ear training and pitch perception shape musicianship.
Audience comments snapshot
Audience comments summary
Public comments are mostly positive and excited about a return to more technical AI discussion, with several people asking for more long-form conversations with researchers and engineers. A few comments focus on the episode’s practical AI/coding angle, and some are lighthearted jokes about the length of the show or Lex’s style.
Comment themes
Technical conversation preference
The comments show strong interest in deep-dive AI discussions rather than broader celebrity or political interviews.
Applied AI enthusiasm
Viewers are engaged with practical applications of AI, especially coding and model-building.
Long-form format as a talking point
The episode’s length is part of the appeal for some viewers and a source of jokes for others.
Audience signals
Demand for more technical AI content
Several commenters explicitly ask for more technical AI episodes and long-form discussions with researchers, engineers, and CTOs.
Hands-on AI and coding interest
One commenter shares a personal workflow involving Python classes, NVIDIA DGX Spark, and building a Llama-based model for business use.
Humor about long runtime
A few comments joke about the episode’s length, including using it as gym motivation or calling it closer to an audiobook than a podcast.
Preference for longer episodes
One commenter says they miss the longer episodes and hope they return.
Representative public comments
Thank you for listening ❤ Check out our sponsors: https://lexfridman.com/sponsors/ep490-sa See below for timestamps, transcript, and to give feedback, submit questions, contact Lex, etc. 0:00 - Introduction 1:57 - China vs US: Who wins the AI race? 10:38 - ChatGPT vs Claude vs Gemini vs Grok: Who is winning? 21:38 -...
Hellllll yeah, back to the AI podcast days. Please do more technical conversations. Interviewing public figures and politicians might be great for reach, I get that. I’ve been listening since your very first podcasts and my favourite have always been conversations like these. More researchers, engineers, CTOs and th...
I started working with AI last year by getting an NVIDIA DGX spark. My sons and I took Python coding classes over last summer. We are developing a Llama model for my business to submit data to multiple platforms we work on. It is going well and has been a learning experience.
I have a deal with myself at the gym - I can’t get off the treadmill until the video ends. Accidentally clicked this video, and now my legs have fallen off
Lex always pushes an audiobook instead of a podcast. 😹
really miss the long episodes, hope you bring them back someday.
Use Crawlora's YouTube comments API with the video and transcript endpoints to collect viewer language, thread activity, and audience signals.