Robot Labor Costs $2 or $35 an Hour — and Trillions Ride on It

By Olivia Hart

A robot costs either $2 an hour, $10 to $12, or more than $35 — depending on which source you believe. A Roland Berger study published earlier this year put running costs at $2 an hour. JPMorgan estimated $10 to $12. Reports on early industrial deployments came in above $35.

Those numbers do not merely disagree. They expose the question that will decide whether physical AI stays a collection of pilots or becomes a labor market measured in trillions of dollars: how much useful work can a robot reliably perform.

Enterprises do not need robots that are cheap in theory, the analysis argues. They need dependable labor in the messy conditions of a factory, warehouse, or distribution center.

What does the $2-an-hour figure actually cover?

The math itself is simple. Divide a robot's purchase or lease cost by the productive hours it works over its life, then add power, maintenance, software, integration, and supervision.

The variable that changes everything is productive time.

Roland Berger's $2 an hour covers energy, software, network, and scheduled maintenance. It excludes the purchase or lease of the machine. Against average U.S. warehouse wages, that $2 only wins if the robots run 60 to 70 percent of the time.

Take a $90,000 industrial robot amortized over five years. At 2,000 working hours per year, hardware contributes about $9 to each operating hour. Add maintenance and other expenses, and the total approaches $12. Cut the schedule to 1,000 hours a year, and the hardware cost per productive hour doubles. Same robot. Same price. Different schedule.

JPMorgan estimates it currently takes about two humanoids to match the output of one human worker — which doubles the effective hourly cost. Productivity per hour improves with software; cost per hour improves only with hours. The productivity problem is being solved in labs. Utilization has to be solved on the floor.

Takeaway: the sticker price matters, but the robot's schedule matters more.

What does Agility's SEC filing reveal?

The clearest public evidence comes from Agility Robotics' S-4, filed ahead of its merger with Churchill Capital Corp XI.

• Digit robots have logged more than 65,000 hours across nine customer sites, including GXO, Schaeffler, and Toyota Motor Manufacturing Canada.
• 2025 net sales: $1.78 million.
• Operating loss: $140.2 million.
• Milestone-contingent orders from a single customer: $300 million.

The filing also lays out Agility's own cost model for its next robot — roughly $400,000 over a five-year life. At one shift of 2,000 hours a year, that works out to about $40 an hour. Two shifts bring it close to $20. Because Agility's fully loaded worker benchmark is $30 an hour, the robot becomes cheaper only after it runs more than 2,700 hours a year — well beyond a single shift.

The $300 million in contingent orders signals real demand. The $140.2 million operating loss against $1.78 million in sales signals how far the business still has to travel.

Why do lab demos keep failing on factory floors?

High utilization cannot rescue a robot that fails whenever conditions change.

Robotics has produced impressive demonstrations for decades: folding shirts, making coffee, moving boxes in controlled environments. Factories and warehouses are different. Objects arrive in unexpected positions. Packaging changes. A person steps into an aisle. Equipment moves. A task that looked simple in a demo becomes a stream of exceptions.

A robot that requires an engineer every time its environment changes is not autonomous labor, the analysis argues. It is an ongoing integration project.

Takeaway: winners will be measured by unglamorous questions — consistent useful work, task changes without reintegration, and enough hours to deliver an attractive cost per completed job.

Is selling robots even the right business model?

Under a Labor-as-a-Service model, the customer buys an outcome: pallets moved, trailers unloaded, parts delivered, packages sorted. The provider assumes responsibility for deployment, maintenance, software, and uptime.

That structure forces vendors to confront utilization directly. A provider paid for completed work cannot hide behind a successful pilot or an ambitious unit forecast.

For customers, the honest comparison is not a robot's theoretical hourly cost versus a wage. It is the cost and reliability of the complete service versus the fully loaded cost of the work today — including overtime, turnover, vacancies, safety incidents, and lost output.

The strongest early applications are the dull, dirty, and dangerous tasks that go unfilled and contribute to injuries. The author notes a March argument that the winners in physical AI would be the most operationally disciplined companies rather than the most visible — a claim that was intuition then, but one these cost figures now support.

Why start with a wedge, not a universal worker?

The industry's most visible ambition is a general-purpose humanoid. That vision may arrive eventually, but the analysis calls a narrow wedge more credible for near-term scale: a recurring problem in an environment structured enough for reliable operation, yet complex enough to create a defensible advantage.

Warehousing, automotive manufacturing, food processing, and industrial material movement fit — not because they make the best videos, but because the work is repetitive, demand persists, and facilities run multiple shifts.

A company might start by automating pallet movement alone. Reliable performance there generates something more valuable than a demo: real-world data. Deployments expose systems to variations no laboratory reproduces. Better models handle more exceptions. Greater reliability opens adjacent tasks. More tasks raise utilization and lower effective cost — how a narrow wedge becomes a platform.

The global market for manual labor is measured in tens of trillions of dollars, and capturing even a small share would create enormous companies. But the opportunity will not be won by putting more robots into pilots. It will be won, the analysis concludes, by turning machines into dependable labor capacity — measured not by how many robots companies sell, but by how much useful work those robots complete.

Disclosure: Navin Chaddha is Managing Partner at Mayfield, an investor in Rhoda AI, a robot foundation model company. Mayfield holds no other physical AI positions.