A not-too-technical overview from MarketWatch, July 1:
Jensen Huang has taken to calling robotics and physical AI the next trillion-dollar opportunity for Nvidia and the market takes the company’s CEO at his word. Nvidia’s physical-AI revenue has run past $9 billion over the trailing 12 months, up from $6 billion the year before, and analysts now treat robots as its second act.
Nvidia’s ambition is to do for robotics what its CUDA platform did for accelerated computing. Huang has called humanoid robots a “multitrillion-dollar economic opportunity.” Nvidia’s newly announced Halos for Robotics safety stack sharpens the point: The company is building the software, compute and safety layer around humanoids, not trying to own the entire machine. Nvidia wants the operating layer underneath — and if physical AI scales the way factory automation has, it will get it.That’s the bull case — and it holds. The catch is the timing. The revenue Huang points to is years from mattering; most of today’s physical-AI money is self-driving and factory automation rather than humanoids. Nvidia just folded the whole business segment into a new “edge computing” line where it cannot be sized on its own. For scale, fiscal 2026 brought in $215.9 billion in total revenue, with data center at $193.7 billion and the line that bundled automotive and robotics together at $2.3 billion, almost all of it self-driving.
The platform thesis is not yet where the cash is. That cash now is showing up one layer down, in the companies Nvidia is pulling into its ecosystem: motion, sensing, power and factory-automation suppliers that already sell to real customers. Right now, Nvidia is the architecture bet. Humanoid-robot makers are the lottery tickets. The cleaner trade is in incumbents that can ride the robotics build-out without needing the humanoid future to arrive on schedule.
The case for the suppliers rests on a number the consultancy McKinsey put in print in April: Actuators account for 40% to 60% of a humanoid robot’s bill of materials, with sensing another 10% to 20% and compute 10% to 15%.
The single most valuable component sits in one of the least developed supplier ecosystems in the stack. Value is concentrated exactly where capacity is thinnest. For investors, that makes the first question simple: Does the company sell into a bottleneck, or into the broader humanoid dream? The closer a supplier sits to motion, sensing, safety or factory integration, the more likely it is to see revenue sooner. The farther it sits from those bottlenecks, the more it depends on the full robot story arriving on schedule.
That is where the contracts are landing. In May, German motion-technology group Schaeffler agreed to deploy up to 2,000 robots from U.K. startup Humanoid across its plants by 2032. It also became Humanoid’s preferred supplier for more than half its joint-actuator demand through 2031, a deal expected to cover a seven-digit number of units, meaning at least 1 million. One robot rollout converted into a recurring component order book. Schaeffler itself reckons actuators are about half the bill of materials in many humanoids, which tells you where it wants to sit....
....MUCH MORE
