English
Subscribe

Editor's notes

Robot Money Splits After the Sale

We counted all 373 players on the Robotics Industry Map. Among unit sellers, 75% in the US and 37% in Korea have money coming in after the sale, yet 15 players that had it still disappeared.

Robot Money Splits After the Sale

Robot stories usually start with unit counts. Look at the companies' money, though, and the wider gap lies in what comes in after one unit is sold. This post follows that single question through the Robotics Industry Map, which covers 373 players across 6 regions, with research as of October 28, 2026.

This time we did not pick standout cases. We counted the whole map by revenue tag, and two things came out. Among companies that sell units, the share with a money route after the sale is far apart: 75% in the US, 70% in Japan and 37% in Korea. But having that route does not keep a company alive: 15 of the 33 players that disappeared had it too.

The robots that sell most are not the ones that keep paying

  • Unit counts and the shape of the money go separate ways. In 2025, about 34.2M consumer service robots and 8,400 surgical robots were sold (IFR ✔). Yet surgical robot maker Intuitive gets 84% of its $10.06B FY2025 revenue from recurring sources, with instruments and accessories alone at about 60% (✔). Instruments work out to about $1,800 per procedure. That reads as a business paid each time the robot is used.
  • Vacuums can sell in volume without revenue and profit moving together. Shipment leader Roborock's 2025 revenue rose +56.5% to RMB 18.695B while net profit fell -31.0% to RMB 1.363B (reported △). First-generation leader iRobot had filter and brush consumables, yet revenue fell from $1.18B (2022) to $680M (2024) (✔), and it passed through Chapter 11 to its contract manufacturer Picea. Having consumables alone does not seem to make recurring revenue solid.
  • Factory robots sit in between. FANUC's robot segment revenue is ¥378.6B (+14.9% ✔), or 44% of the company, and company-wide service revenue has grown 5 years in a row, taking its share from 12.8% to 16.5%. Operating margin, however, fell from 40.8% in FY2014 to 21.4% in FY2025; the periods differ, so the two can't be linked directly, but service share alone can't explain the margin.

Count the whole map and Korea comes out low

  • It is 177 of 285. Of the 373 players, 285 carry the "Hardware sales" tag, and 177 of them (62%) have at least one more tag for money after the sale. In this post, money after the sale means subscription · leasing, maintenance · consumables, software, model API, and apps · content. The other 108 stay on the unit-sales side according to the revenue sources listed on the map.
  • Regions differ widely. Among unit sellers, the share with an after-sale tag is 75% in the US (36 of 48), 72% in the rest of the world (23 of 32), 70% in Japan (30 of 43), 68% in Europe (34 of 50), 56% in China (37 of 66) and 37% in Korea (17 of 46).
  • Japan stands out on maintenance, Korea on installation. In Japan, 19 of 43 carry a maintenance · parts · consumables tag, against 5 of 46 in Korea. Korea instead has more SI · integration tags, 9 of 46 against 3 of 48 in the US. That reads as money closer to delivery and installation than to money that keeps coming after the install.
  • A gap remains even among the same product. Counting only companies that sell cobots, 10 of 12 in Europe, 6 of 7 in Japan and 3 of 7 in Korea carry an after-sale tag. So product mix alone does not explain the gap.
  • There are also reasons not to take these numbers at face value. A tag shows that a revenue source exists, not its share. Korea's rows average 1.02 sources per player, fewer than the US at 1.44 and Japan at 1.31, so Korea may be under-captured. Yet China averages 0.84 and still comes out at 56%, so source count alone does not explain it either. This map cannot tell a real structural difference from a difference in research depth.

Put a few Korean companies side by side and a rule appears, then breaks

  • The cobot-sales side carries big losses. Doosan Robotics has revenue of KRW 33B and an operating loss of KRW 59.5B, Neuromeka a revenue forecast of KRW 19.7B and an estimated operating loss of KRW 18.5B, and Hanwha Robotics revenue of about KRW 10.1B and a net loss of about KRW 13.2B (all △). Rainbow Robotics, where cobot sales are more than half of revenue, has revenue of KRW 34.1B (△); the map lists no profit figure.
  • Profits show up where a maintenance · consumables tag exists. HD Hyundai Robotics has revenue of KRW 263B and operating profit of KRW 21.8B (△), Curexo revenue of KRW 74.5B and a swing to operating profit (✔), and ROBOTIS revenue of KRW 38.94B and operating profit of KRW 3.35B (✔).
  • But a counterexample is right there. SBB Tech, with the same tag, has revenue of KRW 7.199B and an operating loss of KRW 6.8B (△). The three above also differ in size and business. Most of ROBOTIS's revenue is actuators (reported), and Curexo's includes medical device distribution. So it cannot be said that a maintenance tag creates profit. All that can be said is that we did not find, on the map, a Korean company that is cobot-led and profitable.

As unit prices fall, the way of charging changes

  • Unit prices are moving down. Unitree's G1 was cut from $16,000 at launch to $13,500, and the R1 AIR is $4,900 (✔). Cobot prices are also falling under Chinese pressure (△). The map records this as falling per-unit margins.
  • The price to buy is giving way to the price to use. Of Intuitive's 1,721 new da Vinci placements in 2025, 872 were leases, 496 of them usage-based (✔). 1X NEO is $20,000 to buy or $499/month to subscribe, and LOVOT 3.0 is ¥599,500 plus a mandatory ¥9,900~19,800 monthly service fee (✔).
  • Money that comes later carries risks that come later. When Moxie's company closed in 2024-12, the servers shut down and units already sold stopped working (△). Intuitive has also flagged a 3~5% drop in instrument revenue per procedure from H1 2027 (✔). The map calls this the first time it trims the razor-blade price.

Subscriptions did not save them

  • The 33 that disappeared split in two. The map has 8 bankruptcies, 17 discontinued, and 8 absorbed after an acquisition. Of these 33, 16 list only hardware sales and 15 carry an after-sale money tag. Having a route for money after the sale does not guarantee survival.
  • Some added subscriptions. Pepper paired unit sales with a 36-month contract and a monthly subscription but was discontinued, and the serving-robot rental Deliplate was discontinued too (△). Zebra withdrew from the Fetch Robotics business, which even ran RaaS, and put the cost at up to $80M (reported △). Set against Intuitive, whose money arrives every time a surgery happens, the question that comes first reads as whether there is a reason to keep using the robot, not the form of charging.
  • Where players disappeared also differs by form. Wheeled · tabletop robots lost 13 of 53, articulated arms 2 of 53 and cobots 2 of 44. Since the disappeared players are only confirmed cases, these values should not be read as failure rates.

Company values are not yet set in order of later money

  • A business with the later money in place sold for 2.36x revenue. ABB Robotics combines units with service, parts and software, and had FY2025 revenue of $2.331B. Its agreed sale price to SoftBank is $5.375B (✔ · not closed as of October 28).
  • Larger values went to those disclosing almost no revenue. Figure AI is valued at $39B (company announcement ✔), and Skild AI at $14B+ with ARR just past $100M (✔). The map says of these "brain" companies that almost no revenue is disclosed. ABB's price reads as closer to today's money, theirs to future money. Whether those hopes turn into money after the sale is something the map does not answer.
  • Counting shows how much weight sits on expected value. Of the 38 players in the top tier S, 22 reached it on valuation and 16 on revenue. Of 51 humanoid players, 6 have revenue as their key figure, 11 have valuation, and 29 are unconfirmed. The after-sale tag share for humanoids is also 41% (21 players), below 83% for surgical · medical devices (25 of 30).
  • Selling units is not in itself a weakness. Selling mainly units, Unitree made 2025 revenue of RMB 1.699B and net profit of RMB 278M (✔). So the answer here is not "units don't make money" but "at the same unit count, what comes after the sale shapes the company."

Five questions to ask about a robot company

  • What percentage of revenue is money after the sale? A tag and a share are different things. Few companies disclose a share the way Intuitive does with its 84%.
  • When does that money arrive? Per use, at breakdown, and when the calendar turns are all different kinds of money.
  • What happens to the product if the company stops? That is the risk Moxie showed.
  • Does the margin hold when unit prices fall? As at FANUC, margin can move on its own even when service share grows.
  • Is the company's value based on revenue or on expectation? That is why ABB and Figure AI should not be put on the same line.

How to read the numbers

  • Verdicts. ✔ = filing or original source, △ = press report, estimate or research body, ✘ = not the case. Of the 373, 49 have a ✔ key figure and 324 a △.
  • Counted values. Ratios and player counts are values counted from the map data on this post's own basis. The grouping of money after the sale is this post's own, not the map's classification. SI, education, government contracts and investment stakes are left out; including them would change the ratios.
  • Calculated values. The $1,800 per procedure is instruments and accessories revenue of $6.02B divided by about 3.3M procedures. FANUC's China service belongs to a joint venture and sits outside consolidation.
  • What not to compare. The $18.6B global industrial robot installation value (✔) counts units only and excludes software, peripherals and SI, so the money after the sale followed here isn't in that figure. The IFR service robot count behind 34.2M and 8,400 is a sample of 238 suppliers. Annual shipments and cumulative installs shouldn't be set side by side either.
  • Scale tiers. S $1B+ · A $100M+ · B $10M+ · C $1M+. Listed companies use robotics segment revenue and private startups use valuation, so the same S can mean earned money or expected value. The 248 entries in tier "?" have no key figure because they are private or do not disclose robotics segment figures separately.

Unit counts tell you the size of the industry, not whether the money comes back.

Open the Robotics Industry Map