Latest / Minds, Bodies, and Terawatts / Physical Intelligence Raises $1B in Funding

Transcript

Physical Intelligence Raises $1B in Funding · Minds, Bodies, and Terawatts

• 0:00
  [Host] This is Minds, Bodies, and Terawatts, where the money talks and the robots walk. TechCrunch reported on March 27th, 2026 that Physical Intelligence just closed a billion-dollar Series C round, doubling its valuation to five point six billion dollars. The round was led by Radical Ventures, with participation from Jeff Bezos, Thrive Capital, Lux Capital, and returning backer Khosla Ventures. A billion dollars for a company with about 80 employees and no shipping product.
• 0:27
  [Guest] That's roughly twelve and a half million dollars per employee. For context, when OpenAI raised its first billion, it had a product millions of people used daily. Physical Intelligence has research demos and a pitch that says "ChatGPT, but for robots." Investors are paying for a future that doesn't exist yet.
• 0:46
  [Host] So what specifically is the company claiming it can do? Because "general-purpose robot AI" can mean almost anything.
• 0:53
  [Guest] Physical Intelligence published a system called Pi-zero in late 2024, a vision-language-action model they claim can control multiple robot form factors from a single neural network. In their demos, the same model folded laundry, packed boxes, and bused tables across different robot bodies. CEO Karol Hausman has said the goal is zero-shot generalization, meaning a robot encountering a new task or new environment performs it without retraining. That's what the billion dollars is funding: scaling Pi-zero from lab demo to commercial deployment.
• 1:24
  [Host] Zero-shot generalization across different hardware. That's an extraordinary claim. Is there peer-reviewed evidence backing it up, or are we in pitch-deck territory?
• 1:32
  [Guest] Somewhere in between. The Pi-zero paper showed strong results on benchmark manipulation tasks, and the team's pedigree is real — Hausman came from Google DeepMind's robotics division, with co-founders behind key foundation model papers at Stanford and Berkeley. But benchmarks in controlled settings and reliable performance in a messy warehouse are very different problems. The investor thesis is essentially: the people who invented this subfield are now commercializing it, and we'd rather own a piece of that than bet against it.
• 2:03
  [Host] Let's talk about what five point six billion in valuation means for the broader landscape. Figure AI raised over a billion at two point six billion valuation. Tesla's dumping capital into Optimus. 1X has commercial pre-orders. Where does Physical Intelligence fit?
• 2:18
  [Guest] It reshuffles the deck. Figure, Tesla, 1X are all vertically integrated. They build the body and train the brain. Physical Intelligence is saying the brain should be a separate product, sold to everyone — if they succeed, they become the unavoidable supplier; if they fail, they've burned a billion dollars on a theory.
• 2:38
  [Host] Google tried the horizontal play with Android for robotics through various acquisitions and never cracked it. What's different here?
• 2:45
  [Guest] Scale of data and compute. Physical Intelligence told TechCrunch they plan to use the new funding to build what they called the largest robot learning dataset ever assembled, combining real-world demonstration data with massive simulation. Google's earlier attempts didn't have access to today's foundation model architectures or the NVIDIA GPU clusters to train them. Whether that's a sufficient difference or just a more expensive version of the same problem remains genuinely open.
• 3:12
  [Host] OK, a pre-revenue company at five point six billion. Is this genuine confidence or speculative bubble?
• 3:18
  [Guest] Probably both. The underlying technology has made real progress in the last eighteen months. But a billion dollars is a staggering amount for pre-revenue in any sector. The Bezos check is the tell — he runs the world's largest warehouse operation, and if Physical Intelligence delivers a general-purpose robot brain, Amazon is the obvious first customer at scale.
• 3:40
  [Host] That's a detail worth sitting with. Bezos isn't investing as a futurist. He's investing as a buyer.
• 3:47
  [Guest] Which reframes the entire round. Radical Ventures, Thrive Capital, Lux Capital, these are conviction bets on a technology category. Bezos is a strategic pre-purchase. He wants optionality on the software that might run the next generation of Amazon's robotic fleet.
• 4:02
  [Host] What does this mean for companies already selling robots? If you're Figure or 1X, is Physical Intelligence a potential supplier or a threat commoditizing your hardware?
• 4:11
  [Guest] Both. If Physical Intelligence builds a model that makes any cheap robot body perform like a premium one, hardware margins collapse and the intelligence layer captures the value. That's the smartphone pattern — Samsung makes the hardware, but Google and Apple capture the profit. Physical Intelligence is explicitly pitching that outcome for robotics.
• 4:31
  [Host] And there's a workforce dimension the coverage mostly skipped. A plug-and-play AI brain compresses the deployment timeline for industries that have been slow adopters. Small manufacturers, regional warehouses, agricultural operations that couldn't afford robotics engineers.
• 4:44
  [Guest] That's the real acceleration mechanism. The barrier isn't the cost of the robot — it's the cost of integration, the custom programming, the specialized talent. If Physical Intelligence delivers on zero-shot generalization, a warehouse manager could configure a robot the way you configure a new laptop, dropping the adoption threshold from Fortune 500 companies to any business with a loading dock. Workers in food processing, small-batch manufacturing, and last-mile logistics face a compressed timeline they haven't been warned about.
• 5:14
  [Host] There's a piece on unscarcity dot ai called "Humanoid Robots in 2025: Costs, Companies, and Timeline" that maps the current economics of robot deployment. One relevant detail: 1X's deal with EQT to deploy ten thousand robots across three hundred portfolio companies works because cost per hour dropped below three dollars. Physical Intelligence's play could push that lower by eliminating the expensive custom integration step entirely.
• 5:39
  [Guest] And a companion article called "AI Layoffs 2025 to 2030" tracks how companies simultaneously cut human headcount while pouring capital into automation infrastructure — Physical Intelligence's round fits that pattern exactly. The money from Bezos and these venture firms isn't speculative in the traditional sense; it's pre-purchasing the tools companies plan to use as labor substitutes. A general-purpose robot brain could compress adoption timelines that current workforce projections assume will play out gradually.
• 6:06
  [Host] So a billion dollars walks into a company with 80 people and no product. The real story is what it tells us about how fast the industry is about to move, and who's already positioning to buy.
• 6:20
  [Guest] Physical Intelligence might succeed or flame out. But Bezos, Radical Ventures, Khosla, and Thrive Capital all writing checks at five point six billion tells you where the smart money expects value to concentrate. Patrick Déglon calls this the convergence of Mind and Body in his book Unscarcity, the moment when AI cognition and physical robotics stop being separate industries and become one integrated system. That convergence just got a billion dollars closer.
• 6:47
  [Host] You can explore over a hundred research articles, listen to the audiobook, and read the full book online at unscarcity dot ai. This has been Minds, Bodies, and Terawatts.