We begin with the machine itself — what it consists of, how it is sold, and where profit accumulates along its chain — then turn to stocks through the EL7 score. The result is that our quantitative model, a general model that knows nothing about robotics, ranks the sector exactly as this research does.
This research is written to be read from the beginning. The first two sections explain the industry from scratch: what a robot is mechanically, what makes up its cost, how it is actually bought, and where profit accumulates along its chain. Anyone stopping at the end of Section Two will understand the industry, even without reading a word about stocks.
The first structural fact. A robot is essentially a set of joints, and each joint contains a motor, reducer, encoder and sensor. An industrial arm requires six joints, while a humanoid robot requires roughly thirty — about five times the precision-gear content. Motion-control assemblies account for more than 30% of the cost of a fully equipped robot, and more than 50% of the cost of a basic robot.
The second structural fact, which governs every discussion of pricing. The robot itself accounts for about one-third of what the buyer pays. According to the International Federation of Robotics, robot sales were 16.5 billion dollars in 2018, while the turnover of their systems was 48.0 billion dollars. The remaining two-thirds is integration engineering, which does not follow the semiconductor cost curve. Thus, cutting the robot's price to zero reduces the deployment bill by only one-third.
Profit therefore accumulates in the component layer, not the machine layer. We measured margins from our data: Keyence, which sells sensors, achieves a gross margin of 83.7% and an operating margin of 52.1%; Fanuc, which makes arms, achieves 38.4% and 21.8%; and ATS, a pure systems integrator, achieves 6.6% operating. The reason is structural: the component supplier sells a specification approved for ten years, while the robot maker sells a configuration repriced on every project. The average industrial-robot price fell from 31,100 dollars in 2018 to 25,600 dollars in 2024.
When we apply the EL7 score to the sector, our model says the same thing — and it knows nothing about robotics. The median score for robotics specialists is 12.7, versus 84.0 for component suppliers — a gap of 71.3 points. Keyence leads with a score of 97.2, while Serve Robotics stands at 2.3 and ranks 177 out of 185 in its industry.
The factor breakdown reveals the mechanism. Growth is the only factor in which specialists outperform, and they do so clearly: 87.5 versus 67.0. Quality is 29.7 versus 72.9, and valuation 9.0 versus 24.4. The model sees growth as entirely real, just as enthusiasts do — but sees everything beneath it as broken.
The market confirms this in prices. Based on our data through the 21 August close: the specialist basket has fallen 39.2% year to date, the component-supplier basket has risen 34.1%, and the S&P 500 has risen 12.3%. Specialists peaked in October 2025, ten and a half months ago.
As for economics, one comparison settles it. At the price observed today, a humanoid robot recoups its cost in 2.4 years in the United States and 16.1 years in China. A conventional industrial arm recoups it in 7 months and 1.9 years respectively — outperforming in every market we measured by between 3.6 and 8.4 times.
A free account gives you 4 full research papers every month. Or subscribe to EL7 Pro for full access to more than 300 papers.
$700 per year · or $99 per month