Gonçalo Esteves / building capable systems Field Notes →

The Vault / Episodes

Episode

Moonshots 2026-11-16: The Company Outcompeting China and Tesla in Robotics (Brett Adcock, Figure AI)

status · captured added 17 Nov 2026 updated 17 Nov 2026 aired 16 Nov 2026 roboticsaicomputeproductsspace watch the source episode → see it in the graph →

Index card. The material lives at L1: the research archive (transcript, metadata, the research archive). Show: Moonshots with Peter Diamandis · host Peter Diamandis · Dave Blundin · guest Brett Adcock (Figure AI, Hark) · Alexander Wissner-Gross, Salim Ismail, Emad Mostaque absent · recorded on site at Figure, about 10–11 Nov 2026.

Video https://www.youtube.com/watch?v=E864U582FVY
Runtime 2:12:13
Captions YouTube auto-captions. "HARKC" = Hark, "Baku/botq" = BotQ, "in scale" = Nscale, "Danielle Aruse" = Daniela Rus, "Unry" = Unitree, "Turnis" = John Ternus, "Johnny IV" = Jony Ive
Decomposed 2026-11-17. Four sponsor reads. Peter's Figure stake said at 1:40:35, the first time since EP201

Why it is here

The annual site visit to Figure AI, billed since EP303. Three things changed against earlier tape. (1) Brett Adcock's own home timeline moved. In January it was "by the end of 2026… unsupervised multi-day tasks in homes they've never seen before". Here it is "sub six months… I obviously can't promise", tasks "over like hours and hours", and consumer sale "27 28". (2) The guest contradicts two panel calls: GPT-6 Astra-type models are "not real models for robotics" (against Alexander Wissner-Gross on EP311), and "I bet Hark is bigger than Figure in sub six months" (against Alex's Figure-buys-Hark prediction). (3) Peter's Figure stake comes back, at minute 100. The title says "outcompeting China and Tesla"; Brett says China is "certainly" the closest competitor and he hasn't seen much from Tesla "in two or three years". Also on tape: Helix 3 (unreleased), Figure 4's hand, BotQ "over 40,000" units a year, the Nscale deal (✅ $3.5B initial, 100,000 Vera Rubin GPUs, H2 2027), "September 2026, the world ran out of compute", and "We are not ready" for space. One number fails verification: "low tens of millions of dollars per hour to put a human in space"; NASA's published crew-time price is $130,000 an hour. → Moonshots: the disclosures ledger · Unitree · OpenAI

Entities cited

Reading rule. Anything marked unverified or on-air is a claim made on the podcast and has not been checked against a primary source yet. It stays marked until it has. Sourced numbers link to where they came from.