Episode Summary
Executive Summary: Kara Swisher interviews Nathan Myhrvold about his wide-ranging work in science, invention, and food. He argues that curiosity-driven, experimental thinking can generate practical breakthroughs in superconductivity, cancer treatment, chips, energy, climate research, asteroids, dinosaurs, and cooking. Across the conversation, he defends techno-optimism while warning that cybersecurity is a bigger short-term risk than AI.
Main Topics: Curiosity-driven polymathism: Myhrvold explains his approach as broad, experimental, and evidence-based, moving between science, engineering, food, and paleontology with a mindset of testing ideas rather than accepting authority. AI and the open-ended language breakthrough: He frames modern AI as a major scientific advance, especially because large language models showed that sufficiently large data and compute can produce useful language behavior, though not human-like understanding. Cybersecurity as the real near-term AI-era risk: While acknowledging AI concerns, he says digital infrastructure and border security are far more vulnerable now, and that cyber threats are more urgent than existential AI scenarios. Nuclear power, energy demand, and climate technology: Myhrvold argues that future energy demand will require large-scale nuclear power, with renewables insufficient alone, and that geoengineering and carbon removal deserve serious research. Deep science, cancer therapy, and materials innovation: He highlights projects at Intellectual Ventures and TerraPower, including room-temperature superconductors, new lithography methods, and targeted alpha therapy using actinium-225 for cancer. Asteroids and planetary defense: He revisits his asteroid work, emphasizing the Vera Rubin Observatory’s potential to discover millions of asteroids and improve Earth’s preparedness for low-probability, high-impact impacts. Modernist Cuisine and the democratization of food knowledge: Myhrvold discusses how food culture, especially through social platforms, has made experimentation and higher standards in cooking more mainstream, with ongoing work on pastry and bread.
Key Arguments: The most valuable scientific work starts with curiosity, humility, and testing claims against reality rather than relying on status or authority. OpenAI’s language-model breakthrough was historically important because it showed that scale and data could unlock language behavior without hard-coded linguistic rules. Cybersecurity is more urgent than AI because more of society is connected to vulnerable digital systems and the internet is effectively a porous border. The world will need vastly more primary energy over the next 50 years, and nuclear power is essential because wind and solar are intermittent and batteries remain inadequate. Climate change should motivate both emissions reduction and research into geoengineering and carbon removal, because if conditions worsen, society may need technological fallback options. Technological progress usually creates more benefits than harms, and historical fears about job loss or disruption often underestimate how systems adapt and new forms of work emerge. Food culture improves when people demand better food; tools like TikTok accelerate the spread of techniques and experimentation in kitchens.
Data Points: Number of degrees: 3 - Myhrvold’s academic credentials: PhD in theoretical and mathematical physics, master’s in economics, and master’s in geophysics/space physics. Age of college entry: 14 - He enrolled in college unusually early before later studying at Cambridge with Stephen Hawking. Years since Microsoft CTO era: 1990s - Swisher recalls meeting him when he was Microsoft’s chief technology officer. Potential actinium-225 doses per year: 300,000 - TerraPower engineers believe they can produce this many doses annually in a couple of years for targeted alpha therapy. Current actinium-225 supply: Tiny number of doses per year - He stresses that the present global supply is far too limited for large-scale treatment. Moore’s Law improvement: Multiple factors of a million - He describes historical chip price-performance gains driven by semiconductor scaling. Energy demand growth: At least 5x in 50 years - He predicts the world will need roughly five times as much primary energy over the next half-century. Global per-capita energy use gap: 4–5x less than the U.S. - He notes the world average energy usage per person is far below the American level but is rising. German renewables spending: $300 billion - He cites Germany’s long-term renewable investment while arguing emissions still increased after nuclear shutdowns. Climate measurement site: Mauna Kea - He references long-running CO2 monitoring as a key indicator of atmospheric trends. Pet vet bill threshold: Over $1,000 every 6 seconds - This appears in a sponsor read for Fetch Pet Insurance, not the interview itself. Podcast live event date: July 16 - Swisher promotes a live taping in Washington, D.C. with Gina Raimondo.
Pivotal Quotes: "I think cybersecurity is a much more dangerous thing in the short run than any AI issue." — Nathan Myhrvold: He argues that digital vulnerability is the immediate policy and security concern, more than speculative AI risk. "The proposition: if only we spend $10 billion more, it'll work, is a crazy proposition. It is. But god damn it, it worked." — Nathan Myhrvold: On the historical breakthrough behind large language models and why scale-based AI mattered scientifically. "I go in and I try to learn as much as I possibly can. And then I try to test that knowledge by saying, well, if that's true, then we ought to be able to do this." — Nathan Myhrvold: He explains his method for approaching unfamiliar fields across science, engineering, and cooking.
Implications: Listeners get a case for disciplined techno-optimism: pursue ambitious science, but focus policy on cybersecurity, energy infrastructure, and evidence-based climate research rather than panic or ideology.