What the 2026 robot order numbers actually say.

The Association for Advancing Automation publishes quarterly North American robot order statistics, and they are the closest thing this industry has to a real-time demand signal. The second quarter of 2026 came in at 8,940 units worth 622 million dollars, up 4.3 percent in units and 21.3 percent in order value against the same quarter of 2025.1

First-half totals reached 17,995 units worth 1.166 billion dollars, up 2.0 percent in units and 6.6 percent in value.1

The gap between units and dollars

A 4.3 percent unit increase against a 21.3 percent value increase means the average robot ordered got considerably more expensive. Doing the arithmetic on A3's own figures: Q2 2026 averages about 69,600 dollars per unit, and backing out the year-over-year percentages puts Q2 2025 at roughly 59,800 dollars. That is about a 16 percent rise in average order value in a year.2

There are three plausible explanations and they are not mutually exclusive. Robot prices rose, which tariffs and component costs make credible. The mix shifted toward larger, higher-payload arms. Or buyers are ordering more capable configurations, with more integrated sensing and controls, rather than bare arms.

I lean toward the second and third. The sectors driving growth in the first half were semiconductors, electronics and photonics at 35 percent, life sciences and pharmaceuticals at 32 percent, automotive components at 24 percent, and food and consumer goods at 17 percent.1 Semiconductor and life sciences work skews toward precise, sensor-heavy, cleanroom-rated equipment. That is not where the cheap arms go.

Automotive is no longer the majority

Non-automotive customers accounted for 56 percent of units ordered in the second quarter.1 For an industry whose entire technical culture was formed inside automotive body shops, this is the structurally important number on the page.

It matters because automotive and non-automotive buy differently. An automotive line is engineered once, at enormous cost, and then runs a known part for years. A food plant, an electronics contract manufacturer or a metal fabricator changes what it makes far more often, at lower volume, with less engineering staff to throw at each change. The tooling that industry built for the first buyer does not transfer cleanly to the second.

The majority of North American robot demand now comes from buyers whose production changes faster than the traditional integration model was designed to accommodate.

That is the gap our whole company is aimed at, so treat the emphasis accordingly. But the number is A3's, not ours, and you can check it.

Cobots: one in six robots, one dollar in ten

Collaborative robots reached 2,774 units worth 114 million dollars in the first half, which A3 puts at 15.4 percent of units and 9.8 percent of order value.1 That works out to roughly 41,000 dollars per cobot against about 64,800 dollars for the average robot overall.2

Two readings of that. The optimistic one is that cobots are doing what they were supposed to do, which is get robots into places that could never justify a fenced industrial cell, at a price point that makes a first project survivable. The sober one is that after more than a decade of cobots being the most-discussed category in robotics, they are still under a tenth of the money.

Both are true. The cobot solved the cost of the arm. It did not solve the cost of the deployment, which is where the rest of the budget goes, and that is why the category's share of dollars has not tracked its share of attention.

Density says the constraint is not awareness

The IFR puts the United States at 307 industrial robots per 10,000 manufacturing employees, eighth in the world, against a global average of 132 and a North American regional figure of 204.3

Read that carefully, because it is often reported badly. The United States is more than twice as robot-dense as the world average. American manufacturers are not ignorant of robots and are not refusing to buy them. They are eighth rather than first, behind countries with more concentrated electronics and automotive bases.

If the constraint were awareness, marketing would fix it. It is not awareness. The constraint is that each additional cell in an American high-mix plant costs more to justify than the last one, because the easy applications went first. The remaining applications are the awkward ones: more part numbers, shorter runs, more changeover. That is a deployment-cost problem, not a demand problem.

What I would actually do with this

If you run a non-automotive plant and you have been told robots are for the big automotive guys, the data no longer supports that and has not for a while. You are now in the majority of buyers.

If you are budgeting, do not use the 69,600 dollar average as a project number. That is the value of the robot as ordered. A working cell, with tooling, part presentation, vision, safety and commissioning, typically lands at two to four times the arm. Our cost breakdown walks the line items.

And if you are choosing between a cobot and an industrial arm on price, note that the 24,000 dollar gap in average unit value is real but small next to the integration budget on either side of it. Choose on changeover frequency and guarding, not on the sticker.