Robotics — downstream demand sensor
Wedge #2 of the frontier-risk engine. It does NOT pick robot-maker stocks (a likely price-war arena). It grades the GAP between robot hype and deployed reality, and maps who gets paid either way — the actuator/sensor/compute/power bottlenecks every humanoid maker must buy.
Why this exists — in plain English, and for the kids
This is an honesty engine for the robot boom. Humanoid robots are entering their loudest phase — IPOs, demo videos, “Optimus” headlines. Most of those “sales” are still research units, not robots doing paid work. This engine sorts the theater from the reality: it puts every program on a deployment ladder (backed by cited sources) and tracks the gap between how hyped the field is and how much is actually deployed.
The money angle isn't “guess the winning robot brand” (a likely price war). It's the picks-and-shovels layer — the actuators, sensors, and chips every maker must buy no matter who wins. Every call is frozen with a measurable outcome and graded against a “nothing changed” baseline, so the track record is auditable. Everything here feeds the materials mission: real deployment is tomorrow's material demand, read early.
Robots that walk and use hands are all over the news right now. But here's a secret: most of them are still just doing demos — like a kid who can do one cool trick for the camera but can't actually do their chores yet. This is a truth-detector that gives every robot company a score for how REAL it is, and shows the proof.
And instead of betting on which robot will win, it watches the companies that sell the parts — the motors and sensors every robot needs. In a gold rush, the shovel-seller always gets paid.
This is the honest denominator under the hype: how much of the humanoid field has crossed into disclosed paid deployment. The IPO wave doesn't move this number — cited commercial deployments do.
Warehouse-autonomy fleet-scale anchor — the mature end of physical AI (non-humanoid form factor).
cite: Amazon disclosures: 1M+ mobile robots operating across fulfillment (fleet-scale, disclosed metrics).
Surgical-robotics fleet-scale anchor — AI-assisted autonomy is the growth vector, base is deployed.
cite: Intuitive Surgical filings: ~10k installed systems, millions of annual procedures (fleet-scale).
L3->L4 (2026-07-06 crew cite): recurring paid deployment + reorders across a named multi-customer fleet — the clearest humanoid production deployment. Not L5 (no fleet-wide uptime metrics; commercial launch 2026 pending NVIDIA Halos safety cert).
cite: Agility $2.5B SPAC merger w/ Churchill Capital Corp XI (BusinessWire + SEC 8-K/425, CIK 0002074973, 2026-06-24): $300M+ multi-year committed orders across 30+ customer pipeline, Digit commercially deployed at 9 facilities (Schaeffler, GXO, Toyota Canada, Mercado Libre), 65,000+ operating hours, capacity up to 10k units/yr.
THE calibration anchor: a real L4 in mobile manipulation. Proves the ladder's upper rungs exist.
cite: Boston Dynamics Stretch commercial case-unloading deployments (DHL/Maersk and others; ongoing) — recurring paid deployment.
L2->L3 (2026-07-06 crew cite): disclosed commercial order value (800M yuan) across named industrial customers = paid, disclosed terms. Conservative L3 (delivery just beginning) vs L4.
cite: UBTech (PR Newswire, 2025-11-20): Walker S2 begins MASS PRODUCTION + delivery, accumulated orders >800M yuan (~$112M), deployed at BYD/Geely/FAW-VW/Foxconn/SF Express lines.
Pilot-stage at named automakers/logistics; no disclosed paid deployment yet.
cite: Apptronik pilots with Mercedes-Benz + GXO (units at customer sites, 2025-26; terms not disclosed).
Real sustained third-party production work at a named OEM, but no disclosed commercial payment -> stays L2, not L3. Re-cited 2026-07-06 (crew).
cite: BMW Group press release T0458778EN (2026-06-25): Figure 03 begins production work (logistics sequencing) at Plant Spartanburg, extending the F.02 pilot (30k+ X3 vehicles, 1,250+ hrs). Terms NOT disclosed.
Research/rehab units sold -> L2. Old evidence date -> STALE (needs re-verification).
cite: Fourier GR-series units sold to research/rehab buyers; not disclosed labor deployment.
Hardware IS sold (product revenue) but units are research/testing -> NOT deployment. L2 with caveat.
cite: Unitree STAR IPO approval reporting (Reuters/Caixin 2026): 5,500+ humanoids shipped, but buyers are research/education, not disclosed labor deployment.
Consumer preorders + teleop-assisted demos -> staged. Teleoperation dependence is the watch item.
cite: 1X NEO home-humanoid launch/preorder (teleop-assisted demos; consumer preorders, not deployment).
Demo/research stage; DELIBERATELY left at its old evidence date -> flagged STALE, the differential signal that this program has gone quiet and needs re-verification.
cite: Sanctuary AI Phoenix demos/teleoperation research; no disclosed paid deployment.
High narrative weight; evidence remains staged/internal — the canonical hype-vs-truth gap.
cite: Tesla 2026 quarterly updates + demos: in-house factory testing, no external/paid deployment disclosed.
Operator-seeded WITH named-source citations; the robotics-events crew proposes level-change candidates but a level only MOVES with a cite. Levels set conservatively — shipping research units is NOT deployment.
Research + risk-intelligence, not investment advice, and NOT a list of robot-maker stocks to buy. Froth is computed (see provenance); BOM shares are McKinsey-cited order-of-magnitude ranges; deployment truth mass comes from the citation-gated board.