The $25 robot-hour — when humanoid economics close the loop on the labor-substitution math

Figure 03 is billing BMW $25 per robot-operating-hour. US auto labor runs $45-65/hr fully loaded. The labor-substitution math just resolved in favor of the robots, with capex amortized into the rate. This is the trigger event for accelerating humanoid deployment — and the political consequence that compounds with it.

Public per-hour pricing for industrial humanoid deployment has been the missing data point in the labor-substitution conversation. Until Figure 03 commercial deployment at BMW Spartanburg disclosed roughly $25 per robot-operating-hour, the labor-cost comparison was hypothetical. With the number out, the comparison is concrete: $25/hr robot vs $45-65/hr fully-loaded human labor in US auto manufacturing.

The math working at this price point is the trigger event. Robotics has been promising labor substitution for decades; the actual deployment has been blocked by per-unit capex that made the per-hour economics worse than human labor. Figure's pricing structure amortizes the capex into the operating rate, making the comparison apples-to-apples for the customer. At $25/hr robot vs $45-65/hr human, the rational deployment decision for repetitive industrial tasks is the robot.

The pattern generalizes across material-handling roles in automotive, electronics assembly, and warehouse fulfillment. Unitree G1 at $17,990 on Amazon opens the prosumer tier that previously didn't exist. China holds 85-90% of global humanoid shipments, with Unitree as the volume leader. Tesla Optimus consumer sales targeted end-of-2027. Figure operates in the enterprise tier. The market structure is now visible: enterprise pilots → enterprise mass deployment → prosumer adoption → consumer adoption, with that progression happening across 2026-2028.

The political consequence is the part the AI-policy conversation hasn't engaged with yet. Industrial humanoid deployment at the $25/hr-vs-$45-65/hr math runs into the same political environment as the $600B annual AI capex layoff wave. Both are reallocations from labor opex to capital capex. The white-collar version (AI tools replacing knowledge workers) is the version that's been politically loud through 2025. The blue-collar version (humanoid robots replacing material-handlers) is more visceral, with clearer one-to-one job displacement and more compressed timelines.

The framework question: what does the regulatory architecture for industrial humanoid deployment look like, and is anyone seriously drafting it? The answer in May 2026 is "no, not yet, not at the scope the math implies." EU AI Act has narrow provisions for robotic systems in specific sectors. US has nothing comparable at federal level. The state-by-state response is starting to emerge but is fragmented. The deployment curve is moving faster than the policy response, and the gap is going to widen through 2027.

The throughline: we've covered humanoid robotics quarter by quarter as a capability story. The 2026 shift is to an economics story — the unit economics resolved in favor of deployment, the political economics are about to catch up.

LumiChats — Humanoid Robots 2026: Tesla Optimus vs Figure AI vs Unitree → · Humanoid Press — Humanoid Robots News AI Breakthroughs Robotics Trends →