The parts make money before the robot does. A map of the "picks and shovels" laid out beneath the finished machine.
Part 3 dissected the humanoid into five layers of components. Yet if you quietly turn that anatomy chart over, it becomes, just as it is, an industrial map and a chart of where the money flows. If a certain component is indispensable and only a handful of firms can make it, then the moment robots are mass-produced, money pools precisely at that chokepoint.
Before we begin, however, one thing must be made clear. This article is not a piece urging the purchase or sale of any particular stock, but a map explaining whose shoulders the humanoid industry stands on. The company names that appear are merely examples representing each chokepoint, and any judgment about investing, along with its consequences, rests entirely with the reader.
The first thing that comes to mind is the company that makes the robot itself, the integrator of the finished machine. Yet there is an interesting asymmetry here, for among the major firms we met in Part 2, surprisingly few can be bought directly on the stock market. Tesla is listed, but the humanoid is only one part of its vast business; Figure and Unitree, Apptronik and 1X are private; and Boston Dynamics has been folded in as a subsidiary of Hyundai Motor Group.
And so investors reach once more for an old analogy. When a gold rush broke out, the ones who steadily made money were not those who dug the gold but the merchants who sold the picks, the shovels, and the blue jeans. When it is hard to guess which robot company will be the final winner, the logic runs, a firm that supplies the parts every winner must buy may be the more stable chokepoint — and the figure above shows at a glance where those "picks and shovels" lie.
The precision reducer, which Part 3 called the "costliest handful," is reckoned the thickest chokepoint in the value chain as well. A precision reducer such as the harmonic drive is hard to make, so Japanese firms long effectively led the market, the representatives being Harmonic Drive Systems — which turned the very component's name into its corporate name — and Nabtesco, which makes precision reducers of another kind.
Yet cracks are appearing in this oligopoly. In China, firms such as Leader Harmonious are catching up fast and pulling prices down, and in Korea, companies like SBB Tech and SPG are taking on the challenge of domestic production; and since a single humanoid contains dozens of reducers, one per joint, the moment an age of true mass production arrives, the competition over this small part becomes, in itself, an enormous market.
A chokepoint no less important than the reducer is the artificial-intelligence chip that underpins the robot's "brain." Nvidia in particular is aiming for the de facto standard position, offering both the chips that handle the computation by which a robot sees and judges and the development tools that bundle them into something easy to work with; and since whichever company's robot is sold, such a computing device must go inside its head, this too is a textbook pick-and-shovel.
There are chokepoints elsewhere as well. The company that makes the precision motors that go into the joints, such as Nidec; the firms that make the various sensors that detect posture and force; and the battery company that holds the power to move all of it. Take only the battery, which Part 3 called the real bottleneck: giant battery firms such as CATL, LG Energy Solution and Samsung SDI already occupy one axis of this flow.
Looking at this value chain again through a national lens, the positions of Korea and China are especially interesting. Korea has an array of firms with component strength spanning reducers, motors and batteries — among them Rainbow Robotics, in which Samsung holds a stake — and since Hyundai Motor Group has taken in Boston Dynamics, it holds the potential to aim at the finished product and the components together.
China is more aggressive still. The logic of the war of volume seen in Part 2 applies to components as well, as the country rapidly localizes everything from reducers to motors and batteries within its own supply chain and drives prices down. In the end the humanoid value chain spills beyond a mere competition among companies into a contest at the level of nations, over who holds the component supply chain more deeply and more cheaply — and this is the starting point of the United States, Japan, Korea and China stories that begin in the next part.
Passing through four parts that ran from history to the present and from technology to money, we have looked at how the machine called the humanoid is made and where its value flows. Yet seen from a distance, all of this flow is also one face of a vast competition among a few nations over leadership of a future industry.
Technology and capital, in the end, collide along borders. From the next part we widen the view a step and look at this competition through the eyes of nations. We begin first with the United States, which shakes the board with Optimus and Figure and with enormous capital and artificial intelligence.