Michael Thompson, Board Member at Oklo & Joby on Betting on The Future

Episode Summary

Executive Summary: Michael Thompson argues that reindustrialization and AI depend on abundant, reliable energy, making nuclear power, vertical integration, and physical-world technologies central investment themes. He extends this thesis to Joby Aviation, SpaceX, Aurora, and robotics: technologies that benefit from AI, certification, and infrastructure buildout. A major theme is that regulatory and manufacturing breakthroughs can unlock S-curves, moats, and new financing models.

Main Topics: Energy as the foundation of reindustrialization (Priority: 5/5): Thompson frames energy as the core constraint for AI-driven industrial growth and argues nuclear is the most important solution because it is safer than commonly believed and scalable if modularized. Nuclear modularization and regulatory reform (Priority: 5/5): He describes nuclear as historically stalled by bespoke cost-plus projects and NRC friction, but believes recent executive actions and a more competent Washington are accelerating approvals and market viability. Joby Aviation and eVTOL commercialization (Priority: 5/5): He explains Joby’s aircraft, certification path, charging infrastructure, autonomy stack, and hybrid powertrain roadmap as the basis for a long-term moat and future aviation industry. Vertical integration as a competitive advantage (Priority: 4/5): He argues difficult frontier companies must own the full stack—hardware, software, manufacturing, and infrastructure—citing SpaceX, Tesla, and Joby as examples of multiple businesses in one. AI as an amplifier for physical-world companies (Priority: 4/5): Thompson believes the biggest beneficiaries of AI will be companies that build physical products, because frontier models can recursively improve engineering, manufacturing, and operations. Infrastructure and financing after certification (Priority: 4/5): He says certifying aircraft unlocks real-estate, charging, and JV financing options, similar to how FDA approval de-risks biotech and enables acquisitions or capital formation. Robotics and recursive data flywheels (Priority: 4/5): He highlights Nimble.ai and Aurora as examples of how deploying robots or autonomous systems in real settings generates training data that improves future models and products.

Key Arguments: Energy will be the bottleneck for AI and reindustrialization, so pro-nuclear policy is essential. Nuclear is safer than public perception and could become economic if modularized into mass-produced units. The U.S. nuclear industry’s stagnation reflects regulatory friction: 133 reactors were built before 1978, and only two since. Recent executive orders and a more competent federal approach are materially improving the outlook for advanced nuclear and frontier hardware. Physical-world companies will be the main beneficiaries of AI because models can improve real engineering and manufacturing loops recursively. Joby can build a durable moat through certification, infrastructure, autonomy, and vertical integration across aircraft, controls, and powertrain. Charging infrastructure must be designed for aircraft constraints, so Joby’s ground-based cooling approach is a key technical differentiator. Certification is the major de-risking event that can unlock financing, real-estate partnerships, and industry-standard infrastructure. Frontier companies should be judged on long-term TAM and strategic positioning, not near-term EBITDA multiples. Robotics companies can create data flywheels by forward-deploying products into customer environments, which trains the next generation of models.

Data Points: U.S. reactors built before 1978: 133 - Thompson uses this to show the historical pace of nuclear buildout before NRC-era stagnation. U.S. reactors built since 1978: 2 - Used to argue that regulation has sharply slowed nuclear deployment. Year NRC started: 1974 - He references the regulatory era as a turning point for industry slowdown. Thompson birth year: 1976 - He says recent nuclear policy changes are the biggest thing in his life for the industry. Executive orders signed: 4 - He cites the president’s nuclear-related executive orders as a major industry catalyst. Deadline mentioned for reactors critical: July 4 of this year - He says the administration pushed for reactors to become critical by the 250th anniversary. Time between 0 and 78: 133 reactors built - A second framing of the pre-1978 nuclear buildout pace. Time since 1978: 2 reactors built - A second framing of post-1978 stagnation. Joby aircraft configuration: 6 rotors, fixed wing - He describes the current certifying form factor. Battery energy density: Just over 300 Wh/kg - He cites the current Joby battery pack density when explaining charging limits. Charge-time scaling: 1 hour to 15 minutes = 16x heat - Used to explain why fast charging creates thermal challenges. Vehicle weight comparison: Roughly the curb weight of a Model X - Used to compare Joby’s aircraft weight to a Tesla SUV. Takeoff/landing pad size: 75 by 75 feet - He says this is all that’s needed for vertical takeoff and landing infrastructure. Hydrogen range extension: About 6x - He says a hydrogen-hybrid version could extend range roughly sixfold versus full electric. Autonomy acquisition: X-Wing - He says Joby acquired X-Wing to build autonomy stack technology. Pareto frontier cadence: About every 10 days - He describes how rapidly the leading AI model can change. SpaceX TAM cited: $28.5 trillion - He references SpaceX’s stated market opportunity as part of his long-term thesis. Joby initial founding year: 1993 - He says founder JB started the company’s early version then shelved it due to weak batteries. Joby restart year: 2009 - He says JB relaunched the company after making money in other ventures.

Pivotal Quotes: "Where is all this power going to come from?" — Michael Thompson: Explains why he became interested in nuclear as a prerequisite for AI and industrial growth. "If you could figure out how to modularize it... eventually you're going to hit S-curve." — Michael Thompson: His core thesis on how nuclear can become economically scalable. "One of the biggest beneficiaries and perhaps the biggest beneficiary at the application layer is going to be physical world technologies." — Michael Thompson: Summarizes his belief that AI value will accrue to hardware, manufacturing, and robotics companies.

Implications: Listeners should expect AI-era winners to be energy-rich, vertically integrated, and tied to real-world infrastructure. Regulation, certification, and manufacturing scale—not just software growth—may determine which companies become durable giants.