Robots can look astonishingly cheap or eye-wateringly expensive, depending on what the hourly figure includes. A Fortune commentary argues that the decisive measure is how much useful work a robot reliably completes, and how often it is productively running.
Watch Desk analysis
What happened
Fortune’s article compares estimates ranging from $2 an hour to more than $35. Those figures count different things: the $2 estimate covers running costs at scale, while higher figures may include equipment, integration and the realities of early deployments. They are not competing quotes for the same complete service.
The article’s examples show why utilisation matters. A $90,000 industrial robot spread over five years and 2,000 working hours a year contributes about $9 per operating hour before other expenses. At 1,000 hours a year, that hardware cost per productive hour doubles. The article also cites Agility Robotics’ S-4 filing: its model for a next robot puts the cost at roughly $400,000 over five years, with the robot becoming cheaper than the company’s $30-an-hour fully loaded worker benchmark only above 2,700 operating hours a year.
Why it matters
A demonstration can show that a robot performs a task. The commercial test is whether it can keep doing useful work amid changing packaging, unexpected objects and other factory-floor surprises, without constant engineering support. Low utilisation can turn an attractive spreadsheet into a costly machine waiting for its next assignment.
The article argues that selling completed work through a Labor-as-a-Service model could make that test harder to dodge: providers would take more responsibility for uptime, maintenance and deployment, while customers pay for an outcome. It also makes the case for starting with a narrow, repeatable job rather than expecting a general-purpose robot to do everything at once.
Our read
This is a more useful way to judge physical AI than comparing a robot’s running cost with a worker’s wage and calling it a day. Utilisation, reliability and the full cost of delivering a job belong in the same calculation. The figures are examples from a Fortune commentary, not a universal price list, but the underlying question is refreshingly practical: how much dependable work does the machine actually get done?
What to watch
- Whether robot makers publish utilisation and uptime from real customer deployments.
- How often robots need human intervention or costly re-integration when conditions change.
- Whether service contracts tie payment to completed work rather than robot sales or pilot milestones.
Discussion spark: Should robotics companies be judged mainly on how many machines they sell, or on how much dependable work those machines complete?
Sources and evidence
- Source update (16 October 2026, 13:00 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.