“The Man Taking on Tesla in the Race for Humanoid Robots” — Peter H. Diamandis Moonshots EP #116

Episode summary

Brett Adcock, CEO of Figure AI, gives a facility tour showcasing Figure 02 (their second-generation humanoid robot) and lays out Figure’s thesis: humanoid robots as general-purpose labor that require zero infrastructure changes, powered by the maturing AI ecosystem (imitation learning, reinforcement learning, large-scale training infrastructure). The conversation covers Figure’s rapid hardware iteration strategy, the path to sub-$20K unit costs at volume, the BMW manufacturing trial, and Adcock’s provocative claim that 10 billion humanoid robots could be deployed by 2040.

Key arguments / segments

• [00:03:01] Figure 02 upgrades: 3x CPU/GPU, nearly doubled battery (2.3kWh), all internal wiring, exoskeleton load-bearing shells (aviation-inspired), 6 onboard cameras, 4th-gen hands
• [00:07:00] Rapid iteration philosophy: minimum 3 hardware generations to reach commercial reliability; goal is to make the robot software-limited, not hardware-limited (iPhone 1 vs iPhone 4 analogy)
• [00:11:01] Cost trajectory: at high manufacturing volumes, sub-$20K per unit is achievable; at lease rates comparable to a $20K car (~$100/month), everyone could afford one
• [00:13:00] 10 billion robot market by 2040: 3-5B in workforce, every human owns one for errands/chores/childcare tasks
• [00:15:01] Moral imperative argument: if AGI exists only in digital form, humans become the labor force executing AI decisions — humanoid robots prevent that dystopia
• [00:16:00] Figure sees itself as an AI company that does robotics, not a robotics company
• [00:19:00] Robots building robots: once humanoid manufacturing is possible, costs collapse toward zero and GDP spikes
• [00:23:01] $675M raise at $2.6B valuation from OpenAI, Microsoft, Nvidia, Jeff Bezos; BMW trial went well; Figure 02 described as probably the top humanoid hardware in the world
• [00:25:02] AI maturation made this possible now: deep learning for embodied policies, training infrastructure, imitation/reinforcement learning at scale

Notable claims

• Figure is manufacturing ~1 robot per week in California currently [00:06:00]
• Sub-$20K humanoid robots achievable at sufficient volume [00:12:02]
• 10 billion humanoid robots by 2040 — 3-5B in workforce, rest consumer [00:14:02]
• Cost of goods and services trends toward zero with robot labor + renewable energy [00:19:00]

Guests

• Brett Adcock — CEO and founder of Figure AI. Previously founded Archer Aviation (urban air mobility, IPO’d at $2.7B). Put ~$100M of personal capital into Figure to get started. Building general-purpose humanoid robots with OpenAI integration.

Mapping against Ray Data Co

Figure’s framing of humanoid robots as an abundance engine (goods/services costs trend to zero, GDP trends to infinity) is a strong Sanity Check angle. The moral-imperative argument (robots prevent humans from becoming AGI’s manual labor) is a novel philosophical position worth tracking. The hardware-iteration strategy (3 generations to commercial viability) parallels how most deep-tech companies actually scale. Disclosure: Diamandis’s fund Bold Capital is an investor in Figure.

Related

• humanoid-robots
• abundance
• AI-robotics