Robot Fleet Unit Economics

Utilization and uptime are different problems. A robot that is up but idle looks healthy on a maintenance dashboard and is the most expensive failure mode on this page — the capex is spent, the depreciation runs, and nothing is earned against it.

The dashed line is the monthly depreciation each unit must out-earn to be worth owning. This is the chart a software P&L has no equivalent for.

Blended fleet numbers look acceptable; the archetype split shows 35 units (23% of the fleet, $2.9M of capital) that never repay their hardware inside the 48-month depreciation life. They divide into four mutually exclusive groups, and the fixes are different for each:

18 low-utilization sites still earning positive contribution — the volume was never there; redeploy or recover the unit. 1 reliability failures whose field-service cost exceeds revenue outright — worse than idle, because every month they run they lose money. 5 are both at once, and are the first units to pull out. 11 are adequately utilized and contribution-positive but simply not enough to clear capex in 48 months — a pricing or capex-spec question, not an operations one.

18 + 1 + 5 + 11 = 35. In a software business none of these would be more than low-margin; here all four are capital destruction.

Fleets refresh on hardware generations, so a payback month-count beyond the 48-month life is meaningless — the unit is simply a "no". Failing units are ranked instead by the share of their hardware cost they will recover before end-of-life: a unit at 12% recovery writes off ~88% of its capex — that number sizes the decision, and the two right-hand columns time and price it. A money-loser with 3 months of contract left waits for term-end recovery; one with 20 months left is a renegotiation conversation now. And redeployment only makes sense where the move cost is small against the capex being rescued — $8K to rescue an $84K asset at a busier site is an easy yes; moving a unit that would idle anywhere is not.

Utilization takes roughly 4 months to reach steady state. Judging a unit in month one measures the quality of the install, not the economics of the asset — which is why payback here is projected off the trailing three-month run-rate rather than life-to-date average.

Median unit costs $654/month to service — 10.6% of its own capex per year. The mean is $850/month (12.9% of capex). For context, industrial-equipment maintenance typically runs 5–15% of capital cost annually and high-duty-cycle mobile robots sit at the top of that band, so the median here is inside the range and the mean is above it.

The worst 10% of units carry 33% of all field-service cost. Budgeting from the mean over-provisions the healthy fleet and still under-provisions the tail — the actionable unit is the problem robot, not the average robot, which is also why a dollar figure alone is not enough to judge this line.