Moonshots with Peter Diamandis — Episode #229

Guest: Brett Adcock
Title: Humanoid Run on Neural Net, Autonomous Manufacturing, $50T Market
Date: February 11, 2026

Episode Overview

In this expansive and forward-looking episode, Peter Diamandis visits Figure AI headquarters in San Jose to interview Brett Adcock, founder of Figure AI, and track the astonishing progress in humanoid robotics. The conversation dives deeply into the emergence of general-purpose humanoid robots powered exclusively by neural nets, the radical evolution of autonomous manufacturing, data-driven advances, the coming $50 trillion humanoid robot market, the path to robots building robots, and the societal changes ahead. The episode is peppered with insights on AI, robotics engineering, safety, and the transformative economic and cultural potential of ubiquitous intelligent machines.

Key Themes & Discussion Points

1. Explosive Progress in Humanoid Robotics

• Rapid Advancement: The pace of development in AI and robotics has been exponential, with annual changes that feel like a decade in other sectors.
  • “18 months in AI time, that’s like a decade, dude.” (Dave, 02:36)
• Yearly Reinvention: Every year, the technology, hardware, and business models look radically different.
  • “Year to year, the whole business looks completely different.” (Brett, 00:03; 09:17)

2. Neural Nets: The Core of Robot Intelligence

• Full-Stack Neural Nets: Transition from hundreds of thousands of lines of hand-written code to complete neural network-based control for humanoids.
  • “We removed a majority of all [C code] in the Helix one... today we removed the remaining 109,000 lines of C. There’s all neural net, all neural nets today. That’s the full body.” (Brett, 06:47)
• Why Neural Nets? Coding could not scale to the complexity and adaptability required for humanlike manipulation and movement.
  • “You get unexpected behavior... Both good and bad. But things you could never code.” (Dave, 05:36)

3. Data as the Moat & the Engine

• Fleet Learning & Shared Experience: Data gathered from every individual robot is instantly shared across the fleet, creating a rapid, compounding advantage.
  • “Once one robot learns how to do a task, every robot in the fleet knows it. And humans don’t operate like this.” (Brett, 00:14; 29:29)
• Training Sets: A relentless obsession with acquiring and leveraging the right real-world and synthetic data is essential for developing robust, generalizable neural nets.
  • “If you’re in the neural net game, it’s a data play.” (Brett, 29:01)
• Why Only a Few Winners: The ability to continually accumulate, use, and protect unique robot experience data will make this a winner-take-most market.
  • “I think very few humanoid groups [will succeed]... once one robot learns how to do a task, every robot in the fleet knows it.” (Brett, 29:29)

4. Hardware–Software Co-Design & Vertical Integration

• Design for Neural Networks: Figure’s hardware is built from the ground up to suit the neural nets, not the other way around.
  • “We built really looking at the neural net and we said, how do we fit this into a humanoid robot and what are the best sensors?” (Brett, 10:05)
• Vertical Integration: Figure manufactures almost all vital components in-house—from actuators to hands—to control speed, reliability, cost, and innovation.
  • “It would be great if we could go off and buy motors...but it doesn’t work like that.” (Brett, 43:09)
• Iterative Hardware Development: Each Figure version (1, 2, 3) has quickly improved in beauty, cost, capability, and reliability.
  • “We reduced cost like crazy on Figure 3.” (Brett, 43:06)

5. Figure’s Technology Stack Evolution

• Helix 2 Release: Represents the transition to a 100% neural-architecture robot—including full-body reinforcement learning controllers and integrated tactile sensors.
  • “Today we removed the remaining 109,000 lines of C. There’s all neural net, all neural nets today.” (Brett, 06:47)
• Tactile and Visual Sensing: Robotic hands now have palm cameras and tactile sensors for fine manipulation and occluded tasks.
  • “We now have tactile sensors in every fingertip...as well as palm cameras to understand how we’re grasping items.” (Brett, 28:06)

6. Autonomous Manufacturing and Humans Out of the Loop

• Robots Building Robots: The move toward robots on manufacturing lines will enable scaling to millions of units and accelerate the learning/manufacturing feedback loop.
  • “We will put robots on our Baku lines this year.” (Brett, 30:31)
• End-to-End Automation is Key: True value comes not only from physical automation, but generalization to unseen environments and unbroken autonomy (“long-horizon work”).
  • “What’s impressive is like a full end-to-end robot that is generalizing to an unseen place.” (Brett, 20:09)

7. Market Trajectory, Adoption, and Competition

• Scale Is Everything: The market for general-purpose robots could be as large as half of global GDP—$50–$80 trillion—driven by replacement and augmentation of human labor.
  • “It’s like half a GDP...$50 trillion.” (Peter, 47:44)
• Few Strategic Winners: Despite huge numbers of competitors (especially in China), only a handful (likely under 10) will win due to the data and engineering moats.
  • “Far less than 10.” (Brett, 17:49)
• China vs USA: China is the principal competitive region, especially due to scale and energy in both hardware and entrepreneurship.
  • “We don’t see anybody else besides China as a real competitive threat today.” (Brett, 47:04)
• Role of Big Tech: Apple, Google, Nvidia, Meta—every large tech player is investigating or entering humanoid robots.
  • “Every major group in the world will get in this space. You have to, you have like no choice.” (Brett, 47:44; 48:13)

8. Barriers: Safety, Privacy, and Trust

• Safety as Job One: Physical and cybersecurity robustness, fail-safe architectures, and relentless real-world testing are top priorities before home deployment.
  • “Until I put me and my kids and family on board, it’s not safe enough.” (Brett, 87:44)
• Privacy & Data Security: Robots in the home mean careful management of personal data, with dedicated internal cybersecurity teams.
  • “All this is super important. We have an entire team on cybersecurity here.” (Brett, 76:33)
• Intrinsic & Semantic Safety: Must understand both overt and subtle risks (e.g., knocking over a candle, children’s unpredictable movements).

9. Roadmap & Timeline to the Home and Beyond

• Industrial → Home: Industrial/commercial deployments (with BMW, etc.) harden Figure’s robustness. Home use comes next, then healthcare, elder care, and more.
  • “The home is huge in the end.” (Dave, 67:47)
• Alpha Home Deployments: Figure aims to put robots into real homes for long-horizon testing in 2026, with scaling in following years.
  • “My best guess... by end of the year we will be able to put a robot into an unseen home and be able to do fairly long horizon work.” (Brett, 83:15)
• Millions Then Billions: Manufacturing ramps exponentially—facility for 50,000/year planned; billions of robots is a realistic long-term goal.
  • “Every human should have a humanoid to do all your work, then 5 to 10 billion in the commercial workforce...tens of billions over planet.” (Brett, 69:02)

10. AGI, Embodiment, and the Future of Multimodal Intelligence

• Physical Embodiment for True Intelligence: Embodied AGI must physically interact with the world for persistent memory, tool use, reasoning, and truly general intelligence (“Jarvis”).
  • “I want this thing to...reason, have perfect memory, touch the world both physically and digitally.” (Brett, 58:29)
• Unified Models: The goal is a single omnimodal backbone for perception, action, language, and memory.
  • “We believe this all comes down to one model at the end of the day.” (Brett, 33:47)

Notable Quotes & Memorable Moments

On the Data Moat & Robot Learning

Brett Adcock:

"Once one robot learns how to do a task, every robot in the fleet knows it. And humans don’t operate like this.” (00:14, 29:29)

On the Pace of Change

Dave:

“18 months in AI time, that’s like a decade, dude.” (02:36)

On Replacing Code with Neural Nets

Brett Adcock:

"We removed the remaining 109,000 lines of C. There’s all neural net, all neural nets today. That’s the full body." (06:47)

On the Coming Market

Peter Diamandis:

“It’s like half a GDP...$50 trillion.” (47:44)
Brett Adcock:
"It’s going to feel like 2080 up in here." (47:44)

On Partnership with OpenAI

Brett Adcock:

"Our team just ran circles around them... All this stuff was done internally. And at some point it just didn’t make sense to train other folks on how we basically build AI models internally for embedded systems like a humanoid." (11:08)

On Autonomous Manufacturing

Brett Adcock:

"We will put robots on our Baku lines this year." (30:31)
"At some point, I hope in 24 months where all the robots will build all the robots." (69:24)

On Safety & Trust

Brett Adcock:

"Until I put me and my kids and family on board, it’s not safe enough... When I feel safe enough to have a robot in my home full autonomously, end to end, around all my kids—that’s the point I would trust it." (87:44–88:05)

Key Segments & Timestamps

• Touring Figure Headquarters & the Robots (01:56–04:05, 95:58–102:34)
• Figure's Transition to 100% Neural Nets (06:10–10:22, 26:17–29:29)
• Helix 2 and Technical Breakthroughs (24:27–29:01, 26:10–33:47)
• On Partners, Competition & Industry Consolidation (16:33–20:36, 46:09–48:09)
• Manufacturing Scale & Robot-Made Robots (30:01–31:29, 69:24)
• Data, Fleet Learning & AI Moats (29:29–33:36, 72:01–72:56)
• Market Projection & Social Implications (47:44–52:44, 69:02–73:53)
• Safety, Cybersecurity, and Home Deployment (74:59–77:24, 87:28–89:36)
• Personal Insights — Adcock's Background in Senior Care (54:41–56:35)
• General Purpose Robotics & AGI (58:29–62:11, 72:56–73:53)
• User Q&A (80:16–89:36)
• Demonstrating Robot Hardware Generations (95:58–102:34)

Memorable Lightning Round

• Robots on Stage at Deadmau5 Concert:
  "We had several Figure 2s on stage just jamming. We had them all synced so that synced to the music as it danced, which is really cool.” (Brett, 81:50)

• Personal Metric for Trust in Safety:
  "Until I put my, me and my kids and family on board, it's not safe enough to fly anybody." (Brett, 87:44)

Conclusion: The Humanoid Age is Here

Brett, Peter, and Dave paint a vivid picture of an imminent world brimming with affordable, safe, general-purpose humanoid robots—a world-shaking innovation as profound as the birth of the PC or the mobile phone, but supersized. Figure’s obsessive technical strides, integrated approach, and commitment to neural net learning and safety place them at the vanguard. The social, economic, and ethical implications of this transition—across labor, elder care, safety, privacy, and abundance—form the substrata of a moonshot future being realized year by year.

For More

• Peter’s Humanoids Metatrend report: [Substack link in episode description]
• Figure AI: [Company website]
• Follow Peter on X: https://x.com/PeterDiamandis
• Next episode teaser: Exploring abundance, autonomy, and the ethics of AI-driven societies.