Skip to main content

All In Interview - Jensen Huang - Recap and Notes - March 19 2026

March 19 2026

  • No weekly show only for Trump, Jesus and Jensen 🙂
  • (from NotebookLM + myself)
  • Groq acquisition
  • AI Factory
    • Jensen introduces "Dynamo," which he describes as the operating system for the next industrial revolution's "AI factory".
    • The core technology of Dynamo is "disaggregated inference," a complex processing pipeline that spreads different workloads across heterogeneous computing like GPUs, CPUs, switches, and Groq processors.
    • Jensen states that Nvidia has fundamentally evolved from a GPU company into an AI factory company.
  • Agentic AI
    • The AI industry is moving from simple large language models to "agentic processing," where AI agents use working memory, long-term memory, and utilize tools, which puts heavy demands on storage systems.
    • Because modern data centers must run highly diverse workloads (large models, small models, diffusion, auto-regressive), Nvidia developed the Vera Rubin architecture, effectively expanding their Total Addressable Market (TAM) by significantly increasing the amount of hardware racks required for storage and networking.
  • The Three Computers of AI
    • Jensen outlines that the modern AI problem requires three fundamental types of computers:
    1. The Training Computer: Used to develop and create the AI models.
    2. The Simulation/Evaluation Computer (Omniverse): A virtual gym that obeys the laws of physics where AI (like robots or self-driving cars) can be tested.
    3. The Edge Computer: The computer inside the final product, whether that's a self-driving car, a robot, a teddy bear, or a telecommunications base station.
  • Inference and Data Centers
    • Addressing predictions about inference scaling, Jensen notes that inference computing is not just growing a thousandfold, but is on its way to a one billion-x increase.
    • When asked about competitors building 25−30 billion inference factories compared to Nvidia's $50 billion ones, Jensen argues that buyers should focus on the cost of the tokens generated, not the price of the factory.
    • He claims a $50 billion Nvidia factory will generate the lowest cost tokens globally because it produces 10 times the throughput of competitors. He adds that even if competitor chips were given away for free, they wouldn't be cheap enough if they can't keep up with the state of the technology.
  • CEO Strategy
    • When deciding on company strategy, Jensen actively looks for problems that are "insanely hard to do," have never been done before, and uniquely tap into Nvidia's superpowers. Easy problems attract too many competitors.
    • He identifies "Physical AI" as a $50 trillion industry opportunity that has largely been void of technology until now. Nvidia started this journey 10 years ago, and it is now an exponentially growing, multi-billion dollar business.
    • He predicts that "Digital Biology" is two to five years away from its own "ChatGPT moment," where AI will accurately represent and model the dynamics of genes, proteins, and cells.
  • Open Source
    • Discussing desktop computing and hobbyists, Jensen highlights three major inflection points: Generative AI (chatbots), Reasoning AI (grounded information and thinking), and Agentic AI (doing actual work).
    • He praises open-source agentic frameworks (referencing OpenClaw), calling them a reinvention of computing. These systems function as personal artificial intelligence computers because they natively possess memory, scheduling, resource management, and I/O subsystems.
    • Jensen shares anecdotes of agentic systems replacing entire enterprise software stacks in 90 minutes and completing what would normally be a 7-year PhD research thesis in just 30 minutes on a desktop computer.
  • Employee Productivity and AI ROI
    • Addressing whether AI will generate enough revenue to justify the infrastructure spend, Jensen argues that people will pay heavily for AI that actually gets work done, rather than just answering questions.
    • Jensen views heavy AI token usage as a core metric for employee efficiency; he states that if he pays an engineer $500,000 a year, he expects them to consume at least $250,000 worth of AI tokens to amplify their output.
    • He equates this to modern knowledge workers needing "superhuman abilities," completely removing bottlenecks like "this is too hard" or "this takes too long," leaving only human creativity as the limiting factor.
  • Future Moats
    • Jensen argues that the industry fundamentally needs both proprietary, closed-source models and open-source models.
    • Consumers will use proprietary models (like ChatGPT or Claude) for general intelligence, while industries will rely heavily on open models to capture and control their specialized domain expertise.
    • For software entrepreneurs, he explicitly states that their competitive "moat" will be deep specialization; they should become profound domain experts and connect specialized AI agents to their workflows.
    • Jensen thinks Anthropic will grow even faster than Dario predicted - “every software company will be a value added reseller of Anthropic’s technology”
  • Geopolitics
    • On the topic of regulation, Jensen stresses that policymakers must be educated that AI is simply computer software, not a conscious or alien being.
    • He warns against extreme "doomerism," stating that the greatest national security threat to the U.S. is stalling AI adoption out of fear while other countries advance.
    • Geopolitically, Jensen wants the "American tech stack" to lead globally so the U.S. doesn't lose control of this critical sector like it did with solar or telecommunications.
    • Regarding global supply chains, he advocates for re-industrializing the U.S., diversifying manufacturing (to places like South Korea, Japan, and Europe), and maintaining strong, patient partnerships with Taiwan.
  • Robotics and automation
    • Nvidia's strategy in autonomous driving is to provide the underlying platform (the "Android" of self-driving) rather than building the cars themselves.
    • Jensen predicts that high-functioning humanoid robots are only 3 to 5 cycles (roughly 3 to 5 years) away from becoming reasonable products. He believes robotics will be the greatest unlock for human prosperity and economic mobility, helping to solve massive global labor shortages.
    • On the future of jobs, Jensen acknowledges that while some roles (like pure human driving) will eventually be eliminated, many jobs will simply evolve.
    • He uses the example of radiology: despite early predictions that AI would replace radiologists, the technology made scanning so efficient that demand for radiologists actually skyrocketed.
    • Zero shot genomic modeling - and Friedberg built something in 90 minutes on a Sunday night - the power of these tools is incredible
    • Chauffers of the future will be like mobility assistants - do more than just driving
  • His closing advice to young people is to become deeply expert in using AIs and to focus on language skills, as language has become the ultimate programming language for AI.

Comments

Popular posts from this blog

Mxyztplk

kljfkldsajfks djfsdfklj23kj2kl3j423kl j4l23k jrkl;qjr;lwqej klqwjdklasdjklj234kj23kjsa kljdaskd'asdasd sad w 23 424k 23l4k23lk 4l23 4l32 kweerkwqe23 14 wqd asda d 234 2134 23 42343rldla;dkl;sakdl;kasl;dkas;ldk;asdklkasdklal23l;k4klsaldksaldksl;adk

What's your favourite Iron Maiden song?

Part of the NWOBHM ( ignoramuses can go ahead and google that :P ), Iron Maiden have rocked for a long time, with many brilliant compositions (yes compositions, as intricately crafted as any symphony you philistines - the cadence of mcbrain's drums, the richness of dickinson's voice, steve harris's incomparable lyrics and I will stop before waxing lyrical about the guitars of Dave Murray, Adrian Smith and Janick Gers....man!). So what is your favourite Iron Maiden song? I ask because frankly I don't think I have one, and I doubt if its possible for me to single any one out. (get it -- "single"? I use it while referring to songs? mmmpff :D, k, never mind) (wrt the inaugural blog post which was in keeping with the spirit of this webpage, I thought I might as well clear the air regarding the only thing meaningful in the whole piece - the title )

All In Podcast Episode 223 Recap and Notes April 11 2025

April 11 2025 Sacks is back No Friedberg 😟 Ezra Klein joins the show Podcast of The Ezra Klein show Founder of Vox Abundance book with Derek Thompson Larry Summers joins the show Fmr President of Harvard Govt appointments Trump tariffs and China positioning Wild swings in tariff position Stock market hit Volatility Still lot of tariffs in place Markets have lost ~$4T-$6T Could be worse if you count how workers are hurt If they really didn’t care about Wall St then why were they happy about market going up a couple of days ago? This can reduce demand, increase unemployment, increase prices - which is all bad for people US is trading like an emerging market country right now Zeitgeist around America changing to that of Peron’s Argentina style Authoritarian tendencies, cronyism, etc Counter World is happily embracing 10% tariffs for US China isolated Can’t achieve this by asking nicely - that’s not how things work Stock market is crash o...