On September 24 the International Federation of Robotics (IFR) released World Robotics 2026: factories installed more than 600,000 industrial robots worldwide in 2025 (+11% year-on-year), pushing the global operational stock past 5 million units (+9%) for the first time — double the figure of seven years ago. China alone installed 354,000 robots (+20%), 59% of the global total and nearly 60,000 more than the previous record year.

Every trade outlet ran those headlines. But beneath the numbers, the report contains several signals that matter more to the people who actually live with robots — maintenance departments, shop-floor managers, used-equipment buyers and rental operators. This piece lays the data out, then connects it to what we see on the repair bench.

Industrial robots in warehouse awaiting inspection and refurbishment

The key figures, all in one place

Core numbers from World Robotics 2026 (IFR, published 24 September 2026):

  • Global annual installations: 600,000+ in 2025, +11%; forecast 655,000 in 2026 (+9%) and 806,000 by 2029
  • Global operational stock: 5 million units, +9% — doubled in seven years
  • China: 354,000 installations, +20%, 59% of the world total; domestic brands installed 195,000 units (+15%) for a 55% home-market share (57% in 2024)
  • China by sector: electrical & electronics 96,400 (+16%), automotive 78,900 (+38%), metalworking 78,700 (+44%)
  • United States: almost 38,500 (+12%), overtaking Japan as the world's No. 2 market
  • Japan: 36,219, down 19% — slipping from second to third globally
  • Germany: under 25,000 (-8%), yet still 41% of all EU installations; Italy about 7,800 (-11%)
  • India: roughly 10,500 (+15%), a 27% CAGR since 2020, now the sixth-largest market
  • Humanoids: IFR's companion service-robot report (30 September) puts 2025 global sales of full-size humanoids at about 7,000 units — a rounding error next to 600,000 six-axis installations

The shape of the market in one sentence: Asia — China above all — is surging, the Americas are recovering, Europe and Japan are contracting. Here is what each of those trends means if you own, buy or rent robots.

Five million installed base: the aftermarket is the story

Six hundred thousand installations a year is flow; five million operating robots is the base. And a doubling of the base in seven years means one thing: The huge cohort installed in 2018–2020 is now entering its high-failure window.

From our own work-order mix, robots concentrate three failure families between years five and eight of service:

  • Batteries: encoder and main-board batteries last 2–4 years. FANUC SRVO-062 (BZAL) and SRVO-065 (battery voltage low), ABB 20132/30904 main-battery alarms, KUKA encoder zero-position loss — all classic symptoms of an overdue battery. The battery itself costs pocket change, but swapping it with the controller powered down loses mastering, and re-mastering multiplies both cost and downtime tenfold
  • Gearboxes: RV and harmonic gearboxes have defined grease-change intervals (typically 6,000–12,000 hours). Run past the interval and tooth wear accelerates — first noise, then lost repeatability, finally seizure. Replacing all six gearboxes on a six-axis robot with OEM parts generally runs RMB 50,000–120,000, which is exactly where "repair vs. replace with a used machine" flips
  • Servos and cabling: servo motors are rated around 40,000–60,000 hours (8–10 years on two shifts), but cable dresspacks and counterbalance hoses fail at 3–5 years — faster in welding duty

With IFR forecasting installations to keep climbing (806,000 by 2029), the serviceable fleet only grows. The practical takeaway for maintenance teams: keep a log of every robot's battery change dates and grease services. It is far cheaper than reactive repair. No log yet? Start at the next service visit — one spreadsheet is enough.

Domestic brands at 55%: the used market is changing hands

One easily overlooked number: Chinese domestic brands (Estun, Inovance, EFORT, Siasun and others) installed 195,000 units at home, a 55% share. That was 57% in 2024 — a slight slip, but the second consecutive year above half.

The structural effect on the used-robot market:

  • Fleet turnover is accelerating: factories replacing pre-2015 foreign-brand lines with new domestic cells keep releasing FANUC, ABB, Yaskawa and KUKA machines into the second-hand channel. Pearl River Delta pricing: well-kept 2018–2020 FANUC M-20iA or Yaskawa MA1440 class units trade at RMB 80,000–200,000; pre-2015 machines at 30,000–80,000, priced mainly by run hours
  • For used buys, the big four foreign brands still lead: parts are plentiful (servos, gearboxes, teach pendants all circulate), and independent repair shops know them inside out. Used domestic units (Estun, Inovance) are appearing, priced 30–50% below comparable foreign-brand used machines — but the parts channel and third-party repair support are still early-stage
  • Inspection standards don't change by brand: whichever badge is on the arm, a used-robot check is the same seven moves — verify run hours, measure backlash on every axis, listen for gearbox noise, test e-stop and safety circuits, confirm controller version and licences, inspect the dresspack, run a loaded path test. The full method is in our used-robot inspection guide

A September example from our own bench: a 2019 FANUC M-10iD/12 showing 26,000 run hours, all six axes within backlash tolerance, sold at the mid-point of market. The same model year showing 50,000+ hours with slight backlash on three axes prices about 40% lower. Run hours and measured backlash are the two hardest pricing signals in the used market — harder than any brand premium.

Metalworking +44%: welding cells are the next service wave

The fastest-growing Chinese sector in the data is metalworking: 78,700 installations, +44%. Automotive follows at +38%. Both industries share one trait — they run welding: arc and spot.

Welding is the harshest duty a robot sees: spatter into the torch neck and cabling, unstable shielding gas burning contact tips, wire-feed jams scoring liners, UV and heat ageing the dresspack. A six-axis arm in handling duty can stretch to a 12-month service interval; for welding we recommend a 6-month check focused on three things: gearbox grease condition, dresspack wear, and counterbalance seals (if fitted).

There is a rental angle too. Metalworking demand is seasonal — stainless-steel fabricators around Foshan and Jiangmen typically add welding capacity from September to December. A complete used welding cell (Yaskawa MA1440 or FANUC R-2000iB class plus a Fronius/Aotai power source) rents short-term at RMB 12,000–25,000 per month, versus RMB 400,000–800,000 to buy the cell outright. The full arithmetic is in our previous piece, the real ledger of robot rental.

What the US-Japan-Europe split means for Chinese users

United States +12% to global No. 2, Germany -8%, Japan -19%. Superficially someone else's problem — but two transmission lines run straight back to China:

  • Equipment flows: a 19% contraction in Japan means more Japanese factories release equipment into the second-hand channel — the Yaskawa and FANUC units we source from Japanese auctions these years arrive in notably good condition for exactly this reason. Slower German automotive investment has the same effect on European stock (KUKA, ABB)
  • Where humanoids actually stand: about 7,000 full-size humanoids sold globally in 2025, mostly for R&D and demonstration. The IFR itself flags immature safety standards, high maintenance cost and no proven business case. For the next five years the machines doing real work on shop floors remain six-axis arms, SCARAs and cobots — There is no reason to delay or reshape an automation plan around the humanoid narrative

The bottom line

World Robotics 2026 sends a clear signal: installations keep growing around 9% a year through 2029, the base marches toward 7 million, and China holds 60% of it. For new installations, this is the moment of maximum supply-chain and talent maturity. For installed-base owners, 5 million machines are entering peak service age — log them, change batteries and grease on schedule, and most unplanned downtime simply never happens. For the budget-conscious, the window for quality used foreign-brand machines and whole-cell rentals is wide open. Our three business lines — maintenance & repair, used robot sales and rental — map one-to-one onto those three tracks. Not sure which fits your fleet? Call us with your application and an engineer will run a free assessment.

Sources: IFR World Robotics 2026 (published 24 Sep 2026) and the IFR Service Robots report (30 Sep 2026). Used-robot prices and rental rates are September–October 2026 Pearl River Delta references, quoted case by case.

FAQ

QWhat are the headline numbers of World Robotics 2026?

Global installations exceeded 600,000 units in 2025 (+11%) and the operational stock reached 5 million (+9%). China installed 354,000 (+20%), 59% of the world total, with domestic brands at 55% home share. IFR forecasts 655,000 installations in 2026 and 806,000 by 2029.

QHow often should robot batteries be replaced?

Encoder and main-board batteries typically last 2–4 years (FANUC suggests about 2 years, ABB about 3 — check your controller manual). The critical rule: replace them with the controller powered up, or you lose mastering. Alarms like SRVO-062/065 or ABB 20132/30904 mean the battery is due — don't defer it.

QWith domestic brands past 50% share, are used foreign-brand robots still the better buy?

For now, yes: FANUC, ABB, Yaskawa and KUKA units have the deepest parts circulation, the most mature third-party repair ecosystem and the best residual value. Used domestic robots (Estun, Inovance) cost 30–50% less and can work on a tight budget with local support — but their parts channels are still maturing.

QHow many humanoid robots were sold in 2025? Will they replace six-axis robots?

IFR counts about 7,000 full-size humanoid sales globally in 2025, mostly for R&D and demonstration — against 600,000+ industrial robot installations the same year. Safety standards, maintenance economics and business cases are all immature; the shop-floor workhorse for the next five years remains the traditional industrial robot.

QDo welding robots need shorter service intervals than handling robots?

Yes. Welding duty brings spatter, heat and UV that accelerate dresspack and torch-neck wear — inspect every 6 months versus 12 months for handling duty. Priority checks for a welding cell: gearbox grease condition, dresspack wear, wire feeder and contact tip state.