Robot Orders Tracker: North American Orders by Industry, and the Baseline to Beat
In short: North American companies ordered 10,989 robots worth USD 723.5 million in the third quarter of 2026, 24.8% more units than a year earlier, according to the Association for Advancing Automation (A3). Orders from automotive component suppliers rose 90.4% and from plastics and rubber 67.7% in units, while food and consumer goods fell 23.7% in units. This tracker logs each quarter and turns each industry row into one question for the plant: what does the station you want to automate make today, by shift, and why does it stop?
- Through September 2026: 28,984 robots worth about USD 1.89 billion, up 9.6% in units. Cobots were 1,831 of the third quarter's robots, 16.7%.
- A3's order values work out at about USD 65,839 a robot and USD 34,844 a cobot (order value divided by robots ordered). That is the arm, not the cell.
- Example plant: one assembly station, one cobot cell. Priced on the routing rate, it pays back in 17.0 to 20.7 months. Priced on measured output, with the press that feeds it as the limit, it takes 22.1 to 27.0 months and makes 7,800 parts a week, not 12,600.
Tracker, updated 18 November 2026. Next update: when A3 publishes its fourth-quarter 2026 figures. Sources are listed at the end.
Robot order statistics show where plants are spending on automation. They cannot show whether a given cell will pay. That depends on a record a business case built from cycle times leaves out: what the station makes now, shift by shift, with its stops and their reasons. For automotive suppliers, whose robot orders nearly doubled in the third quarter, it is the same record that sits behind OEE for automotive Tier 1 suppliers: good parts against takt, with the losses known by machine and shift. This page logs A3's quarterly figures, newest first, reads each industry row as a plant question, sets A3 beside two other robot statistics, and prices one robot cell two ways for an Example plant.
North American robot orders, quarter by quarter
A3 publishes robot orders by North American companies each quarter, in units and in dollars, as changes on the same period a year earlier. The third-quarter figures below were checked against A3's own release of 28 October 2026. The trade reports that printed A3's releases are linked as well: The Robot Report on the third quarter, DC Velocity's list of the third-quarter industry rows, Modern Materials Handling on the second quarter and The Robot Report on the first quarter.
| A3 figure | Q1 2026 | Q2 2026 | Q3 2026 |
|---|---|---|---|
| Robots ordered | 9,055 | 8,940 | 10,989 |
| Order value | USD 543 million | USD 622 million | USD 723.5 million |
| Change in units on a year earlier | -0.1% | +4.3% | +24.8% |
| Change in value on a year earlier | -6.4% | +21.3% | +26.1% |
| Order value per robot (value divided by units) | USD 59,967 | USD 69,575 | USD 65,839 |
| Cobots ordered, share of units | 1,637, 18.1% | 1,137, 12.7% | 1,831, 16.7% |
| Automotive components | +28.1% units | +20% | +90.4% |
| Automotive OEMs | -35.1% units | first half: -25% | +7.5% |
| Plastics and rubber | +25.2% units | first half: +6% units | +67.7% units |
| Food and consumer goods | +16.0% units | +18% | -23.7% units |
| Reported | 15 May 2026 | 11 August 2026 | 28 October 2026 |
Source: A3's release of 28 October 2026 for the third quarter, and the trade reports linked above for every quarter; "Reported" is the publication date of the report cited. Order value per robot is order value divided by robots ordered.
Four rules for reading the table:
- Industry rows are changes, not counts. The releases give each industry's change on a year earlier, not its number of robots, so the rows show growth and cannot be ranked by size or added up.
- Units or value. Where A3's lists say "units", the table says so. For the automotive rows in the second and third quarters and the second-quarter food row, the wording does not say, so the table leaves them unmarked.
- Value per robot is an average of a mix. It divides the order value of all robots by their number: small cobots and large robots together. It is the value of the robots on order, not the price of a working cell.
- An order is not an installation. A3 counts orders placed by North American companies. Other statistics count robots installed or shipped, as the comparison further down shows.
Q3 2026, reported 28 October 2026: four rows as plant questions
North American companies ordered 10,989 robots worth USD 723.5 million in the third quarter, up 24.8% in units and 26.1% in value on the third quarter of 2025, which A3 called the strongest quarter of the year so far. That took the first nine months to 28,984 robots worth about USD 1.89 billion, up 9.6% in units and 13.3% in value. A3 linked the rise in semiconductors and electronics to investment in US semiconductor manufacturing and demand for electronics supporting AI and other applications, and named higher input costs, trade uncertainty and the timing of large manufacturing projects as factors to watch. Four rows speak directly to discrete, plastics and food plants.
Automotive components: up 90.4%
Orders from automotive component suppliers climbed 90.4% in the third quarter, after +28.1% in units in the first quarter and +20% in the second. Automotive OEMs went the other way in the first half: down 35.1% in units in the first quarter, down 25% over the first half, then up 7.5% in the third quarter, and still 17% below a year earlier through September. A3's first-quarter release, as The Robot Report printed it, put the component rise down to the segment's typical lag behind OEM ordering cycles.
The plant question: if you supply the parts, is the station you want to automate sized for the customer's programme, and what does it make today against that rate?
- Price the cell at the call-off you hold in writing. A3 tied suppliers' orders to a lag behind OEM ordering cycles, while OEM robot orders through September were still 17% below a year earlier (the two A3 rows read together). Test the payback at a lower call-off as well as the programme figure.
- Three rising quarters can mean several stations at once. When a segment's orders grow three quarters running, by 28.1% in units, then 20%, then 90.4%, a supplier may have more than one station out for quotes. Rank them by measured loss hours, not by the order in which the quotes arrive.
- The cell joins a running process. Its rate will be judged against takt and the customer's weekly demand, so the baseline it replaces has to be on file before the cell arrives.
Plastics and rubber: up 67.7% in units
Plastics and rubber orders rose 67.7% in units in the third quarter. In the first quarter they rose 25.2% in units and 32.6% in value; over the first half, 6% in units. Japan points the other way. In the Japan Robot Association's July to September 2026 results, JARA's member companies shipped 175 robots to Japanese plastic products makers against 188 a year earlier, down 6.9%, and 1,132 robots for plastic moulding at home and abroad against 1,327, down 14.7%.
The plant question: does the station you would automate work at its own pace, or at the pace of the press that feeds it?
- A lump, not a trend. A first half of +6% followed by a quarter of +67.7% reads as one segment's orders arriving together, an inference from the two rows. The months between a quote and an order are time to measure.
- The press sets the supply. A cell that assembles or packs moulded parts can only handle what the press moulds. Count the press and the station over the same shifts, as the Example plant below does with press 7.
- North American, not global. JARA's members shipped fewer robots for plastic moulding while North American moulders ordered more. The A3 row describes buyers in one region in one quarter.
Food and consumer goods: down 23.7% in units
Food and consumer goods was the only row in A3's third-quarter list to fall: down 23.7% in units in the third quarter, after +16.0% in units in the first quarter and +18% in the second, and +17% in units over the first half. Other statistics show food automation rising. The International Federation of Robotics (IFR) release on the US market counts 2,900 robots installed in the US food and beverage industry in 2025, up 30%, and JARA's members shipped 275 robots to Japanese food, beverage and tobacco makers in July to September 2026, against 174 a year earlier.
The plant question: if your end-of-line project slipped this year, what did the station make while it waited?
- Read it as timing. One falling quarter after two rising ones, while installations and shipments elsewhere rose, looks like projects moving between quarters, the factor A3 itself named; the reading of the rows is an inference, not A3's. It does not show food plants turning away from robots.
- A delayed order is a measuring window. The months between quote and order are when the manual station's counts, stops and reasons get recorded. For hand-pack stations, the guide to measure manual packing before automating sets out what to count and how.
- Formats change the case. A food line that changes formats several times a shift hands the cell a changeover, and a first-off check, at each one. Those are stops the cell inherits, as the table further down sorts them.
Cobots: 1,831 units, 16.7% of robots ordered
Companies ordered 1,831 collaborative robots worth USD 63.8 million in the third quarter, about 56% more units than a year earlier and 16.7% of all robots ordered, but only about 8.8% of the order value (63.8 / 723.5). Per robot that is USD 63,800,000 / 1,831 = USD 34,844, against USD 659,700,000 / 9,158 = USD 72,035 for the other robots. Through September, cobots were 4,605 units worth about USD 178 million, 15.9% of units and 9.4% of value, and 43.8% of robot orders in life sciences and 36.9% in semiconductors and electronics. Their share of units moved from 18.1% in the first quarter to 12.7% in the second and 16.7% in the third.
The plant question: is the arm the cost you are deciding on, or the cell around it?
- The arm is a part of the price. At A3's averages a cobot is about half the order value of another robot. In the Example plant below, the cobot average is 24.9% of the installed cell; the feeder, fixture, conveyor, risk assessment, programming and installation are the rest (Example data).
- Collaborative is an application. OSHA's robotics standards page says there are currently no specific OSHA standards for the robotics industry, and lists the consensus standards instead: ISO/TS 15066 (collaborative robot safety), RIA TR R15.806 (testing methods for power and force limited collaborative applications) and RIA TR R15.306 (task-based risk assessment methodology). Ask which ones the cell is built and tested to.
- The type follows the station. Whether a cobot or a guarded robot fits depends on the parts, the speed and the people beside it; the trade-offs are set out in cobot vs traditional robot.
The next quarter is added above this one, with the date of its release.
Q2 and Q1 2026: the releases before
Q2 2026, reported 11 August 2026
North American companies ordered 8,940 robots worth USD 622 million, up 4.3% in units and 21.3% in value. The first half reached 17,995 robots worth USD 1.166 billion, up 2.0% in units and 6.6% in value. In the quarter, semiconductors and electronics rose 38%, automotive components 20%, food and consumer goods and metals 18% each, and life sciences 9%. Non-automotive customers took 56% of the units. Cobots were 1,137 units worth USD 44 million, 12.7% of units and 7.1% of value.
The plant reading: value grew five times faster than units, so the order value per robot rose to about USD 69,575, the highest of the three quarters (622 million / 8,940), in the quarter when the cobot share was lowest. A total that rises faster in dollars than in units can hide a shift to larger robots.
Q1 2026, reported 15 May 2026
North American companies ordered 9,055 robots worth USD 543 million, down 0.1% in units and 6.4% in value. Automotive OEM orders fell 35.1% in units and 48.2% in value. Every other row in the list rose: life sciences +54.1% in units, semiconductors and electronics +31.7%, automotive components +28.1%, plastics and rubber +25.2%, all other industries +24.5% and food and consumer goods +16.0%. Cobots were 1,637 units worth USD 69.8 million, 18.1% of units and 12.9% of value.
The plant reading: a flat total hid a split between one falling customer group and rising ones. A total is no guide to one industry, and an industry row is no guide to one station.
Three robot statistics, three answers by industry
A3 is one record. Two others count robots differently. The IFR's World Robotics 2026 release counts industrial robots installed in each country in a year, and JARA counts orders, production and shipments by its member companies each quarter.
| Record | What it counts | Period | Food | Plastics | Automotive |
|---|---|---|---|---|---|
| A3 | Robot orders by North American companies | Q3 2026 on Q3 2025 | Food and consumer goods: -23.7% in units | Plastics and rubber: +67.7% in units | Components: +90.4%; OEMs: +7.5% |
| IFR | Industrial robots installed in the US | 2025 on 2024 | Food and beverage: 2,900, +30% | Not given in the US release | Automotive industry: 13,500, -1% |
| JARA | Robots shipped by JARA members to customers in Japan | July to September 2026 on 2025 | Food, beverages and tobacco: 174 to 275, +58.0% | Plastic products: 188 to 175, -6.9% | Auto parts: 1,691 to 1,839, +8.8% |
Sources: the A3 reports linked above, the IFR's US release dated 24 September 2026, and JARA's July to September 2026 results (file dated 19 October 2026). Percentages for JARA's rows are calculated from its unit counts and match the changes JARA prints.
The totals differ as well. The IFR counts 38,400 robots installed in the US in 2025, up 12%, which put the US ahead of Japan's 36,200 as the second-largest market after China. Its global release counts more than 600,000 installations worldwide, up 11%, an operating stock of 5 million, and forecasts 655,000 installations in 2026, up 9%. For the rest of A3's region, it counts almost 4,000 installations in Canada, up 5%, and fewer than 5,200 in Mexico, down 7%. JARA's members took orders for 67,003 robots in July to September 2026, up 36.1% in units and 46.6% in value, and shipped 50,403, of which 41,226 went abroad. In the US release, the IFR's president said growth came from the food industry and from sectors such as warehousing, logistics and medical, while automotive stayed stable as the largest customer.
None of these contradicts the others: they measure orders, installations and shipments, in different places and periods. Together they show that an industry's row moves with projects and regions. None of them can say whether one plant's station will pay back. That figure comes from the station.
Before the robot: what a new cell inherits from the station
A cell bought to replace a manual station takes the station's place in the flow, and with it some of the station's losses. A business case built on the routing rate for the station and the quoted rate for the cell counts none of them. The rule this guide uses:
Cell good output a week = the smaller of (cell rate x (planned hours minus the stop hours it inherits) x its own yield) and (the good parts the step before supplies).
To use it, sort the station's measured stop hours by what a cell would do to them. The sorting below is a working classification, a starting point to agree with the integrator.
| Loss at the manual station | How it shows in the record | What a cell does to it |
|---|---|---|
| Waiting for parts from the step before | Stop with the reason "waiting for parts", by the step it waits for | Inherits it: the cell waits for the same parts |
| Changeovers and first-off checks upstream | Stops after a mould, tool or format change on the step before | Inherits it, and adds a changeover of its own if the part changes |
| Pace below the routing rate while running | Running rate below the standard cycle | Removes it, and replaces it with the cell's own cycle, which has to be measured at handover |
| Material presentation: empty bins, tangled parts | Stops with "bin empty" or "parts tangled" | Changes it: a feeder takes away the walk to the bin and can add jams of its own |
| Full containers waiting for removal | Stops with "container full" | Changes it: a buffer conveyor holds more, and someone still moves the containers |
| Breaks, meetings and handovers | Planned hours shorter than paid hours | Partly removes it: a cell can run through them if it is kept fed and the step before runs too |
| Rejects caused by the incoming part | Good parts below parts assembled, with the reason | Inherits it: the cell assembles the same faulty parts |
Two counts make the baseline complete: the station's good output by shift, and the output of the step that feeds it. A press or filler with a PLC gives the second. A manual bench without a machine signal needs a tally, a counter or a camera for the first; the methods for assembly benches are in computer vision OEE for manual assembly. How a fast station ends up waiting behind a slower one is explained in starvation and blocking, and the gap between a rated and a real rate in nameplate capacity vs real capacity.
Worked example: one seal-and-housing station, two business cases
Example plant. Every operating figure in this section is Example data; the robot price anchors, the wage and the overtime rule are sourced. The plant is not any company named above.
A plastics and rubber products plant moulds a housing on press 7. Next to the press, an assembler presses a rubber seal into each housing, clips on a cap and places the part in a returnable tote for a Tier 1 customer. The station runs two shifts, five days a week. Each shift is 8 paid hours and 7 planned hours after breaks, a meeting and cleaning: 70 planned and 80 paid hours a week, over 47 production weeks a year. The customer takes 7,700 parts a week today and has asked for 12,000 a week from next year.
Step 1: the proposal. An integrator quotes a cobot cell with a seal bowl feeder, a press-fit fixture and a conveyor that holds four totes, rated at a 20-second cycle, 180 parts an hour, and tended by press 7's operator. The robot is priced from A3's third-quarter averages: USD 34,844 for a cobot and, as a high case for a larger arm, USD 65,839 across all robots (A3's order values divided by units). Integration is USD 105,000 (Example data). The cell costs 34,844 + 105,000 = USD 139,844 to 65,839 + 105,000 = USD 170,839.
Step 2: case A, built from the routing. The routing gives the station a 25-second cycle, 144 parts an hour: 144 x 70 = 10,080 parts a week. The cell at its rating makes 180 x 70 = 12,600. To reach 12,000, the manual station would need 1,920 more parts, 1,920 / 144 = 13.3 hours, so two Saturday shifts of 7 planned hours (2 x 7 x 144 = 2,016 parts). The cell needs none. Case A counts two savings:
- Weekday assembler hours: 2 assemblers x 40 paid hours x 47 weeks = 3,760 hours a year. The BLS Occupational Employment and Wage Statistics for May 2025 give miscellaneous assemblers and fabricators (51-2090) in plastics and rubber products manufacturing (NAICS 326) a mean wage of USD 20.22 an hour, median USD 19.12. 3,760 x 20.22 = USD 76,027.20.
- Saturday overtime avoided: 2 shifts x 8 paid hours x 47 weeks = 752 hours. The US Department of Labor's overtime page says covered, non-exempt employees must receive at least time and one-half for hours over 40 in a workweek, and that Saturday work needs no premium unless it is overtime. Each assembler has already worked 40 hours by Friday, so every Saturday hour is overtime: 752 x 1.5 x 20.22 = USD 22,808.16.
Case A's yearly saving is 76,027.20 + 22,808.16 = USD 98,835.36. Payback: 139,844 / 98,835.36 = 1.4149 years, or 17.0 months, to 170,839 / 98,835.36 = 1.7285 years, or 20.7 months. The case promises 12,600 parts a week.
Step 3: 20 production days of measurement. Before signing, the plant counts the station's good output by shift, and the assembler gives a reason whenever the station stops. Press 7's PLC counts its good housings over the same shifts. Weekly averages:
| Weekly average | Morning shift | Afternoon shift | Both shifts |
|---|---|---|---|
| Good parts | 4,060 (116 a planned hour) | 3,710 (106 a planned hour) | 7,770 |
| Waiting for housings from press 7 | 90 min | 270 min | 360 min |
| Seal bin empty or seals tangled | 45 min | 45 min | 90 min |
| Full tote waiting for removal | 30 min | 30 min | 60 min |
| First-off check after a mould change | 0 min | 30 min | 30 min |
| Stops in all | 165 min | 375 min | 540 min |
| Running hours | 32.25 h | 28.75 h | 61.0 h |
| Good parts per running hour | 125.9 | 129.0 | 127.4 |
Example data. Press 7 made 7,800 good housings a week, 7,800 / 70 = 111.43 a planned hour; the station turned 7,770 of them into good parts, 99.6%.
The afternoon makes 3,710 against the morning's 4,060, but not because the afternoon assembler is slower: while running, the afternoon makes 129.0 parts an hour and the morning 125.9. The difference is press 7, whose mould changes are planned in the afternoon. A cell bought to fix "the slow afternoon" would wait through the same mould changes.
Step 4: sort the stops. Using the table above: waiting for housings (360 minutes) and the first-off check (30 minutes) are inherited, 390 minutes or 6.5 hours a week. The seal bin (90 minutes) and the full totes (60 minutes) are removed by the feeder and the conveyor in the quote, 150 minutes. The cell's capacity in the hours left is 180 x (70 minus 6.5) = 180 x 63.5 = 11,430 parts a week.
Step 5: case B, built from the measurement. The cell can only assemble the housings press 7 makes: 7,800 a week, so the smaller number is 7,800, less whatever the cell rejects once it runs. At 180 an hour the cell needs 7,800 / 180 = 43.3 hours for a week's housings, 61.9% of its 70 planned hours; the rest it waits. The customer's 12,000 is out of reach whoever assembles: press 7 is 12,000 minus 7,800 = 4,200 housings short, which at 111.43 an hour is 37.7 more press hours a week, well beyond two Saturday shifts of 7 hours. So case B counts no Saturday saving, only the weekday assembler hours: USD 76,027.20 a year. Payback: 139,844 / 76,027.20 = 1.8394 years, or 22.1 months, to 170,839 / 76,027.20 = 2.2471 years, or 27.0 months.
| Item | Case A: routing rates | Case B: measured by shift |
|---|---|---|
| Manual station, parts a week | 10,080 (144 x 70) | 7,770 (4,060 + 3,710) |
| Cell, parts a week | 12,600 (180 x 70) | Up to 7,800 (press 7's housings) |
| The customer's 12,000 a week | Met by the cell | Not met by either; press 7 is 4,200 short |
| Weekday assembler hours saved a year | 3,760 | 3,760 |
| Saturday overtime avoided a year | 752 hours | None counted |
| Yearly saving | USD 98,835.36 | USD 76,027.20 |
| Cell cost | USD 139,844 to 170,839 | USD 139,844 to 170,839 |
| Payback | 17.0 to 20.7 months | 22.1 to 27.0 months |
Step 6: what changed. Case A overstated the yearly saving by USD 22,808.16, 23.1% of its own figure, and the payback moved out by 5.1 to 6.2 months (from the unrounded years: 1.8394 minus 1.4149 and 2.2471 minus 1.7285). The bigger miss is the promise: case A told the customer 12,600 parts a week, and case B shows 7,800 until press 7 changes. The wage is the OEWS mean wage, not a full labour cost with benefits, so both savings are floors; with a loaded labour rate both paybacks shorten, and case B stays the longer of the two.
Step 7: what to do first. Press 7's afternoon mould changes cost the station 270 minutes of waiting and 30 minutes of first-off checks a week. Shorter changes, using SMED (single-minute exchange of die), or a different slot for them, raise the housings press 7 supplies; repeat the 20 days and price case B again. If the plant adds press shifts to chase 12,000, each added shift with the cell needs no assembler: 8 x 1.5 x 20.22 = USD 242.64 for each Saturday shift. Price that saving inside the press plan, because it only exists with the press hours.
Book a demo and see OEE and losses per line and shift.
Automate now, fix the feed first, or wait: choosing
The worked example ends in one plant's answer. The table sets out the options with the anchors from this page.
| Option | Choose it if | Cost anchor on this page | Watch for |
|---|---|---|---|
| Automate the station now | The station's own losses (pace, material, containers) are its largest stop hours, and the step before has parts to spare | Cell USD 139,844 to 170,839 (Example); 22.1 to 27.0 months on measured output | Inherited stops: the cell waits for the same parts |
| Fix the step before first | Waiting for parts is the top stop reason, as press 7's 360 of 540 stop minutes were | Engineering time on changeovers (not priced here) | The gain has to show in the step before's count before the cell is ordered |
| A cobot cell instead of a guarded robot cell | People work beside the station, the parts are light and floor space is short | Robot about USD 34,844 against USD 72,035 for other robots, Q3 2026 averages from A3's totals | The application, its risk assessment and its tested limits decide safety, not the arm |
| Keep the station manual and move the press changeovers | Demand is flat and the gap between shifts comes from the schedule | A changed changeover slot and training time | The afternoon's 106 parts a planned hour against the morning's 116 came from press 7, not the assembler |
| Add hours at both steps | Demand rises and both the station and the step before can staff them | USD 242.64 per Saturday assembler shift (8 x 1.5 x 20.22) | Both steps need the hours, or one waits |
Questions to put to the integrator before the quote is signed:
- How was the quoted cycle measured, and does it include container changes, feeder refills and gripper changes?
- What does the cell do when the step before stops, and how is that wait shown and recorded?
- Which standard is the cell built to, such as ISO 10218-2 for robot systems and system integration, and which risk assessment method is used, such as RIA TR R15.306?
- For a collaborative application, which limits apply under ISO/TS 15066, and how will they be tested, for example under RIA TR R15.806?
- What is the handover test: how many parts, on which shift, at what rate, with a changeover inside it?
- Who restores the cell after a fault on the second shift, with which spares, and how fast?
Before the purchase order: a 20-day measurement plan
Each step produces a number that case B needs. Plant, production and CI managers can split them as the plant runs.
This week
- Write down the station's routing rate, the cell's quoted rate and the hours each one assumes.
- Start counting the station's good output by hour and shift: a tally sheet, a counter or a camera.
- Give the operators a short list of stop reasons, with "waiting for parts" split by the step the station waits for.
Over the next 20 production days
- Count the step before over the same shifts, from its PLC or a counter.
- Total the stop hours by reason and by shift each week, and look for a shift gap that the step before explains.
- Sort the hours into removed, inherited and changed, using the table above.
This quarter
- Work out the cell's output as the smaller of its capacity in the hours left and the step before's supply.
- Price case B at the wage for the occupation and industry in the BLS tables, on the hours the cell really removes.
- If waiting for parts is the top reason, fix the step before and repeat the 20 days.
- Write the case B output into the purchase order as the handover target, by shift.
After start-up: hold the cell to the measured baseline
A cell that starts is a new machine with its own losses. Track it the way the station was tracked: good parts by shift, stops by reason, and the inherited ones kept apart, so the handover is judged against case B and not against the 180-an-hour rating. The team that tends, restarts and adjusts it improves with experience, as the learning curve describes; plot its stop minutes against parts made, not against calendar weeks. When a fault stops the cell, the first steps are in industrial robot fault troubleshooting. For the press and the cell together, OEE for plastics manufacturing shows which losses to watch on moulding lines.
The record behind a robot business case
Every case on this page turns on the same record: good output by station and shift, stops with their reasons, and the output of the step that feeds it, kept long enough to compare before and after. A robot order statistic says where peers are spending. The record says what the spend will buy.
Fabrico is a cloud OEE and light MES platform. It shows OEE and losses by machine, line and shift, with data from the machine's PLC or an IoT device with sensors on the machines you connect, and operators give the reason for a stop at the machine. With the selling price and margin of each SKU entered, losses are priced in money. On manual stations, AI cameras (an add-on) count output, which gives a station a measured count by shift before it is automated. The building blocks are OEE software and a light MES for the production record.
Book a demo and see losses per machine and shift.
What people ask next
How many robots did North American companies order in 2026?
Through September 2026, 28,984 robots valued at about USD 1.89 billion, up 9.6% in units and 13.3% in value on the same months of 2025, according to A3's figures. The third quarter alone was 10,989 robots worth USD 723.5 million, up 24.8% in units and 26.1% in value. This tracker adds the fourth quarter when A3 publishes it.
Which industries increased robot orders in the third quarter of 2026?
On A3's figures: semiconductors and electronics/photonics up 97.6% in units, automotive components up 90.4%, plastics and rubber up 67.7% in units, life sciences up 38.7% in units, metals up 37.0% in units and automotive OEMs up 7.5%. All other industries rose 0.5% and food and consumer goods fell 23.7% in units. The figures are changes on a year earlier, not unit counts by industry, so they show growth, not size.
What does an industrial robot cost compared with a robot cell?
A3's third-quarter 2026 order values work out at about USD 65,839 per robot and USD 34,844 per cobot (calculated from A3's totals), across a mix of robot types and sizes. That is the robot, not the cell: feeders, fixtures, guarding, the risk assessment, programming and installation come on top. In this page's Example plant, integration is USD 105,000 of Example data, so the cobot is about a quarter of the cell.
How do you calculate the payback of a robot cell?
Divide the installed cell cost by the yearly saving the measured baseline supports. Count the labour hours the cell removes at the wage for that occupation and industry, and count extra output only up to what the step before supplies and the customer takes. In the Example plant the same cell pays back in 17.0 to 20.7 months on routing rates and in 22.1 to 27.0 months on measured output.
Are cobots cheaper than traditional industrial robots?
Per robot, on A3's third-quarter 2026 figures: 1,831 cobots were worth USD 63.8 million, about USD 34,844 each, against about USD 72,035 for the other 9,158 robots (calculated from A3's totals). The figures mix sizes and uses, and the arm is only part of a cell's cost. Cobots were 16.7% of the units ordered and about 8.8% of the value.
What should you measure at a manual station before automating it?
Good output by shift, stops with their reasons, and the output of the step that feeds the station, over at least 20 production days. Then sort the stop hours into those a cell removes and those it inherits, such as waiting for parts or first-off checks after a changeover upstream. The cell's output is the smaller of its capacity in the hours left and what the step before supplies.
Sources
Sources accessed on 17 November 2026 unless marked as accessed on another date. Dates are publication dates where a page gives one; other dates are labelled as what they are.
- Association for Advancing Automation (A3), North American Robot Orders Increase 24.8% in Third Quarter of 2026, news release dated 28 October 2026, read in a browser, accessed 18 November 2026
- The Robot Report, North American robot orders climb 24.8% in Q3 of 2026, says A3 (A3 third-quarter 2026 data), 28 October 2026, accessed 17 November 2026
- DC Velocity, North American robot orders increased 24.8% in third quarter (A3 third-quarter 2026 data), 30 October 2026, accessed 17 November 2026
- Modern Materials Handling, Robot orders increase in Q2 (A3 second-quarter and first-half 2026 data), 11 August 2026, accessed 17 November 2026
- The Robot Report, North American robot orders remain flat at the start of 2026 (A3 first-quarter 2026 data), 15 May 2026, accessed 17 November 2026
- International Federation of Robotics, Five Million Robots now Operate in Factories Globally (World Robotics 2026 release), 24 September 2026, accessed 17 November 2026
- International Federation of Robotics, U.S. now Second-Largest Robotics Market, Following China (press release, PDF), 24 September 2026, accessed 17 November 2026
- Japan Robot Association (JARA), July-September 2026 Results: orders, production and shipments of manipulators and robots, JARA member companies (PDF), file dated 19 October 2026, accessed 17 November 2026
- U.S. Bureau of Labor Statistics, Occupational Employment and Wage Statistics, May 2025 national industry-specific estimates: Plastics and Rubber Products Manufacturing (NAICS 326000), Miscellaneous Assemblers and Fabricators (51-2090), May 2025 estimates, released 15 May 2026, accessed 17 November 2026
- U.S. Department of Labor, Wage and Hour Division, Overtime Pay, undated page, accessed 17 November 2026
- Occupational Safety and Health Administration (U.S. Department of Labor), Robotics: Standards, undated page, accessed 17 November 2026

Real Fabrico tiles, sample data.








