Robot Eyes and Brains
harsh✌️@harsh
−7.0%
since Sep 30
The idea
I think robots go from demos to real volume within two to three years, so instead of picking a robot maker I own the parts every robot needs: sensing chips, machine vision, and the processors that do the thinking, plus small slices of pure-play lidar for extra upside. Weights are heaviest in sensing silicon and Nvidia, lighter in vision specialists and lidar, with no single name dominating and some cash set aside to add on dips. I plan to hold for years and would rethink it if robot shipment volumes stall instead of ramping.
What it owns
- Every joint and gripper needs analog chips to measure position, force, and power, a foundational layer in every robot built. · Analog DevicesMakes the analog chips that convert touch, motion, and temperature into signals robots can use, with content in nearly every industrial machine., Texas InstrumentsThe scale producer of analog and embedded chips scattered throughout robots, with dozens of sensors per machine and no single-maker dependency., Allegro MicroSystemsMakes the magnetic sensors that tell motors exactly where they are, and humanoid robots have dozens of motors across their joints., onsemiSells both the image sensors that enable vision and the power chips that drive motors, two essential functions in every robot., Monolithic Power SystemsMakes the chips that manage power inside compact machines, and a battery-powered robot depends on efficient power delivery to function., STMicroelectronicsMakes the motion sensors, gyroscopes, and accelerometers that tell a robot how its body is moving in space., TE ConnectivityMakes the connectors and sensor wiring that tie a robot's parts together, unglamorous content in every single robot built., Sensata TechnologiesMakes the force and pressure sensors a robot needs to grip without crushing and to feel its own load, enabling the sense of touch.40%
- Robots cannot work without cameras and imaging systems; these companies make the sensors and vision processors that enable robot sight. · CognexSells the cameras and software that let machines see in factories, a pure play on machine vision that benefits directly from robot adoption., Sony GroupMakes more image sensors than anyone else, and every robot that ships needs camera eyes regardless of which robot maker wins., NovantaSupplies precision optics and motion components inside robots and automation equipment, a picks-and-shovels way to get paid per robot shipped., Teledyne TechnologiesOwns a deep bench of industrial imaging and specialized cameras that robots need beyond ordinary vision lenses., AmbarellaMakes chips that process camera data right on the robot instead of sending it to the cloud, essential for robots that cannot afford latency.27%
- The processors that run a robot's artificial intelligence on board, where the core computation happens. · NVIDIAIts chips and robotics software form the default foundation that robot makers build on, so the take depends on this platform remaining the industry standard., QualcommIts chips run artificial intelligence on the device with low power, fitting robots that cannot carry server infrastructure., Lattice SemiconductorMakes small reprogrammable chips that handle logic between a robot's sensors and its main processor, core content in the robot's nervous system.17%
- Depth sensing lets robots move safely around people and obstacles, and these companies supply the lidar and 3D sensors that robots need. · Hesai GroupAlready ships lidar at higher volume than rivals and has proven it can manufacture at scale, positioning it well if robotics becomes a second market., OusterSells lidar into warehouses, industrial sites, and robots today, not just car prototypes, with revenue growing across real-world applications., Aeva TechnologiesMakes lidar that measures velocity as well as distance, which helps robots move safely around people in shared spaces.8%
- Cash8%