Robotics-as-a-Service is the default business model pitched to us by roughly 70% of hardware founders in 2025. Most of those models do not survive contact with a spreadsheet. The math is narrower than the pitch, and the companies that scale inside that narrow band look quite different from the companies that do not.
The three numbers that decide the model
Unit economics in RaaS compress into three headline figures: fully-loaded CapEx per unit, realized monthly revenue per unit at steady-state utilization, and annual service cost per unit. Everything else is commentary.
Our heuristic for a viable RaaS model: monthly revenue at steady state should recover the fully-loaded CapEx within 18 months, and annual service cost should not exceed 20% of annual revenue. Companies inside that envelope can scale. Companies outside it are subsidizing customers.
Fully-loaded CapEx is not the BOM
The CapEx number that matters is not the bill of materials. It includes the BOM, assembly labor, freight, import duties, installation, pre-deployment testing, and the first six months of warranty reserve. For most hardware we see, that full number runs 2.2–2.8x the BOM. Founders who model RaaS payback using BOM-only CapEx are off by a factor of roughly two-and-a-half on their break-even point.
Utilization curves bend later than you think
In every RaaS pitch we have seen, the utilization curve ramps to 85% in month three. In every RaaS deployment we have actually observed, the utilization curve ramps to 40–50% in month three and reaches 75% somewhere between months nine and fourteen. That gap kills cash flow.
We burned eighteen months of runway before we realized we had built our financial model on the average utilization of a mature fleet — not the utilization curve of a new customer ramping up.
What the ramp actually looks like
- Months 1–2 — customer operations team learning the robot, usage under 25%
- Months 3–5 — shift supervisors integrating robot into planning, usage climbs to 45–55%
- Months 6–9 — formal KPIs attached, usage stabilizes around 65%
- Months 9–14 — usage climbs to contractual target, typically 75–80%
- Month 15+ — steady state, fleet operates at designed utilization
The service cost trap
Early RaaS deployments almost always have service costs above 30% of revenue, often well above. Founders assume this will compress as the product matures. It does, but more slowly than planned: in our dataset, service cost reached the 20% target on average 26 months after first deployment, not the 12 months typically modeled.
The implication is simple and uncomfortable. A RaaS company needs to raise enough capital to survive a 26-month service-cost compression. That is usually 40–60% more than the initial financial model suggests.
What good looks like
The RaaS companies that reach scale share a few characteristics. They price the first year of service as a separate line item rather than bundling it into the monthly fee. They instrument the fleet obsessively from day one so that the engineering team can debug field issues without a truck roll. And they build predictive maintenance into firmware before they have ten units in the field.
A defensible model for 2026
A RaaS business that we would underwrite in 2026 looks like this: $85K fully-loaded CapEx per unit, $6.5K monthly revenue at 75% utilization, 24% year-one service cost falling to 18% by year three, and a fleet that reaches 75% utilization by month ten on average. Every lever around that is a dial you can turn, but the basic shape has to hold.
Everything else — the growth rate of the fleet, the product-led expansion within accounts, the move from single-customer deployments to multi-customer micro-fulfillment — is layered on top of that foundation. Teams that get the unit economics wrong at the core are not going to grow their way out of the problem.