← Cross-cutsCross-cut · Robotics

Embodied AI, from demo to monopoly

Four Teardown studies cover robotics and embodied AI — Unitree, Intuitive Surgical, Neuralink and John Deere. Read together, they span the entire commercialization arc, from a pre-revenue moonshot to a proven monopoly — and they agree on one thing: the durable moat is never the robot, it's the installed base and the switching cost around it.

Four stages of the same journey

The cluster is best read as a timeline. Neuralink is the earliest stage — 21 human implants, pre-revenue, a binary late-2020s regulatory path; striking demos, no business yet. Unitree is mid-commercialization — shipping legged robots at volume with revenue up 335% to ¥1.7B, but on hardware, with the software-defined business still ahead. John Deere is an incumbent retrofitting a century-old product into autonomy (See & Spray, self-driving tractors) to build a per-acre software annuity. And Intuitive Surgical is the destination — a two-decade monopoly with ~11,400 da Vinci systems installed and ~86% recurring revenue. Placed side by side, they show what each earlier stage is trying to become.

Intuitive is what 'won' looks like — and why

Intuitive Surgical is the cluster's reference point because it proves where the value actually sits. It earns ~$10B at a premium multiple not from selling robots but from the razor-and-blade model around them: every da Vinci system locks a hospital into years of high-margin instruments, service and training, so ~86% of revenue recurs. The robot is the customer-acquisition cost; the installed base is the business. That is the template every other company here is chasing — and the reason Intuitive's own risk isn't a better robot but the first credible rivals finally arriving to contest a monopoly that switching costs, not technology, protected.

The moat is the lock-in, not the hardware

Read the four together and the pattern is unmistakable: hardware leadership is necessary but never sufficient. Unitree has radical cost leadership (>90% in-house parts, ~60% gross margin) — but its own central question is whether cheap hardware becomes a durable, software-defined business, because hardware alone commoditizes. John Deere's bet is explicitly the annuity: ~500M acres and recurring revenue toward ~10% by 2030, defended by data and dealer lock-in, not by the tractor. Even Neuralink's eventual moat, if it gets one, is a vertically integrated per-procedure model. The cluster's clearest lesson: in robotics you win by owning the installed base and the switching cost, exactly as Intuitive does — the robot is just how you get there.

Capital intensity and regulation gate every one

What the four also share is that the gate is rarely the engineering — it's capital and regulation. Neuralink's entire value rests on an FDA path that is binary and years away, against less-invasive competitors. Intuitive's rivals must clear the same regulatory and clinical bar that took it two decades. John Deere rides a deep farm downturn and a right-to-repair fight. Unitree faces US security and geopolitical scrutiny on top of the software gap. Embodied AI is capital-intensive and safety-regulated by nature, which is itself a moat for whoever clears the bar first — and a multi-year wall for everyone behind them.

Where they agree — and where they split

All four agree that embodied AI is inflecting and that the durable advantage is the installed base plus switching cost, not the device. They split on stage and proof: Intuitive has the model fully built and must now defend it; John Deere is converting an incumbency into an annuity; Unitree must turn a hardware cost-lead into software economics; Neuralink must convert demos into an approved, reimbursable product at all. The demand for robots and autonomy isn't the question. Which of these reaches Intuitive's razor-and-blade endpoint — and which stalls as a hardware company or a science project — is.

The cluster at a glance

CompanyRobot / domainModel & moatStage / scaleBiggest risk
Unitree RoboticsPrivateQuadruped dogs + humanoids>90% in-house parts · ~60% GM¥1.708B rev · +335% YoYSoftware gap + US security/geopolitics
Intuitive SurgicalNASDAQ:ISRGda Vinci soft-tissue surgical robot + IonRazor-and-blade · ~86% recurring moat~11,400 systems · $10.06B rev (+21%)First real rivals + ~50x multiple
NeuralinkPrivateImplantable brain-computer interface (N1)Vertically integrated, per-procedure B2B2C21 implanted, pre-revenue feasibilityBinary late-2020s FDA/regulatory path
Deere &NYSE:DEAg autonomy · See & Spray, 9RXPer-acre SaaS · data + dealer lock-in~500M acres · ~10% recurring by 2030Deep ag downturn + right-to-repair

Figures as of each study’s stated date (2026-06); on mixed bases and not strictly comparable. See each teardown for sourcing and the full weighing.

The four studies — and the question each turns on

Unitree Robotics (宇树科技)PrivateDoes radical hardware cost-leadership translate into a durable, software-defined business as embodied AI inflects or deflates?Read the full weighing →Intuitive Surgical, Inc.NASDAQ:ISRGCan a two-decade installed-base-and-switching-cost monopoly hold its premium as the first credible rivals, GLP-1 demand shifts, and a de-rated ~50x multiple arrive at once?Read the full weighing →Neuralink Corp.PrivateCan a feasibility-stage, pre-revenue BCI convert striking demos into an approved, reimbursable product before less-invasive rivals — and has its ~$9B mark outrun the evidence?Read the full weighing →Deere & Company (John Deere)NYSE:DEIs the precision-ag and recurring-revenue pivot a durable moat, or mostly a way to defend margins through a deep farm downturn and a right-to-repair fight?Read the full weighing →

This is the kind of reading the Desk does for you

A cross-cut takes four teardowns and one thesis and asks what they say together. The Desk does the same for the companies you actually own — your thesis, the rivals that move it, and the tripwires that would change your mind.

See the Desk →