The arm is getting cheaper. The cell is not.
A lender on a robot fleet is lending on a falling price curve for the metal and a rising one for the work it does. This report separates the two, and says what resells.
Built from the Safewaterline robot arms dossier, September 2026, about 250 sourced URLs. Estimates are marked. No Safewaterline loan data is used.
The short version
- The Chinese six axis average selling price fell from about ¥200k to ¥120k between 2020 and 2025. A Borunte 5 kg arm resells for $2,000 to $3,100. Fairino lists a CE marked 10 kg cobot at $10,199 against a UR12e at $44,636. Imported arm resale value follows the Chinese floor down.
- Installations are at records: 542,076 industrial robots in 2024, a preliminary 621,000 in 2025. China is 54% of the world and Chinese brands are 57% of China. IFR.
- No independent cheap arm maker has been profitable. Borunte ran a negative gross margin in 2023. Dobot lost ¥84M in 2025 at a 46% gross margin. The margin sits with the integrator and the dealer, at 40 to 47%.
- Fixtures, grippers, guarding and integration are half or more of a cell's cost and resell for close to nothing. The arm resells. The cell does not.
- Robots as a service exists but is thin. Formic in the US, a ¥1B Chinese leasing market, no dedicated industrial arm lessor in India. The financing slot is open and that is where fleet level risk data is needed.
The price floor
Reducers are about 35% of an arm's bill of materials and Chinese reducer makers already sell at 35 to 60% of Nabtesco and Harmonic Drive prices at 23 to 37% gross margins. An illustrative 10 kg six axis bill of materials is ¥35k to ¥52k. Branded Japanese and European arms carry list prices three to five times that. For a lender the implication is direct: the gap between a UR, FANUC or KUKA list price and the Chinese floor is brand and support, and that premium decays faster in resale than the hardware does.
Estimate. The desk's working assumption is that a branded cobot loses 45 to 55% of list in year one on the secondary market and a Chinese arm 60 to 70%. No published comp series exists. This is the first parameter a robot curve will need.
Who buys arms, and where the fleets are
Handling is 51% of installs, welding 15%, assembly 10%. Cobots were 64,500 units in 2024, 12% of the market. The fastest growing lines are the ones that tend machines: China's metal and machinery segment grew 31% and is 90% Chinese branded. India installed 9,120 units, 45% automotive, with parts suppliers up 40%. IFR China, IFR India.
Fleets a lender will see: machine tending cobot fleets at contract manufacturers, injection moulding take out robots at plastics plants, welding cells at auto Tier 2 suppliers, and mobile robot fleets in warehouses. Each has a different resale depth and a different enforcement problem, which is why the risk score is per fleet type.
What resells
| Component | Share of cell cost | Resale |
|---|---|---|
| Arm and controller, branded | 30 to 45% | resells through dealers and used machinery brokers; the only part with a market |
| Arm and controller, Chinese | 15 to 25% | resells at or below new Chinese price; thin outside Asia |
| Gripper, tooling, fixtures | 15 to 25% | job specific; near zero |
| Guarding, conveyors, cell frame | 10 to 20% | scrap or re-use on site; near zero |
| Integration, programming, commissioning | 20 to 35% | none |
Shares are desk estimates from integrator quotes in the dossier, not comps.
Mobile robot fleets add a software lock: the fleet manager is the vendor's, so the pool of buyers is the vendor's other customers. That is the resale depth score of 10 on the risk panel.
Financing today
Formic sells a working cell for a fixed monthly fee and owns the hardware, with about $60M of equity plus equipment debt. Rapid Robotics' $2,100 a month version is defunct. Chinese leasing is about ¥1B in 2025 with six axis rents of ¥5k to ¥30k a month. In India SIDBI machinery loans run 8.8 to 10.5%, NBFC loans up to 24%, and no industrial arm lessor exists. A ₹8 lakh cell on a 36 month lease at 14% costs about ₹27k a month against about ₹68k a month for three shifts of operators. Formic, SIDBI.
The lessor, not the arm maker, will capture that spread. The lessor's book is the fleet a Safewaterline lender will be asked to refinance.
The lender's questions, and what the desk has
| Question | What exists | What the desk needs before a curve |
|---|---|---|
| What is a 3 year old UR10e worth? | dealer asks, no transaction series | 20 comps per model from the desk's own sales or a broker partner |
| How much of the loan is in the arm? | integrator quotes, itemised | a standard split on every appraisal |
| Can the arm be pulled and resold? | serial plates, controller logs | site access rights in the loan, as for GPUs |
| Does the cell keep earning after a change of owner? | utilisation from the controller | a monitor on the controller, the robot page on the capital site shows one |
Until the first column fills, robots have a framework and a risk score on this portal, not a curve.
Sources
IFR World Robotics 2025 · Manufacturing Dive, 2025 preliminary · IFR China · IFR India · China six axis ASP · Leaderdrive reducers · Formic RaaS · SIDBI machinery loans · Standard Bots Series C · Safewaterline robot arms dossier, 2026-09-21, four research passes.