Robotics Is Having Its Smartphone Moment
— By Sarah Beckwith
The convergence of improved actuator hardware, foundation models for embodied AI, and declining manufacturing costs is producing a robotics capability inflection point that resembles the 2007 smartphone transition more closely than any previous robotics wave.
Key Implications
- Boston Dynamics, Figure AI, Physical Intelligence, and Apptronik are all shipping commercial humanoid robots in 2025–2026. The question has shifted from 'can humanoid robots work in unstructured environments?' to 'which verticals will reach cost-effectiveness first?'
- Foundation models for robotics — embodied AI trained on video and sensor data rather than text — are solving the generalization problem that made previous robotics waves brittle. Robots that can interpret novel instructions and handle unexpected situations are qualitatively different from rule-based automation.
- Warehouse logistics, light manufacturing assembly, and last-mile delivery are the near-term deployment verticals with the clearest unit economics. Healthcare, elder care, and general-purpose household assistance are the medium-term opportunities with the largest total addressable markets.
- The smartphone parallel extends to platform dynamics: the companies that build the robotics operating system layer — the iOS/Android equivalent for embodied AI — will capture disproportionate value as hardware commoditizes.
The robotics industry has had many 'revolutionary moments' that didn't quite arrive. Industrial robots have been transforming manufacturing since the 1970s. Drone delivery has been 'five years away' for a decade. Self-driving has repeatedly missed its timelines. So when I say that robotics is having its smartphone moment, I want to be precise about what I mean — and why I think this time is different.
The smartphone transition wasn't just about the device. It was about the convergence of several independent technology trajectories arriving at the same time: capacitive touchscreens, lithium polymer batteries, mobile processors, always-on connectivity, GPS, and the App Store distribution model. Any one of those elements in isolation would have produced incrementally better phones. Together, they produced a qualitative transition.
Robotics is at a similar convergence point, and the elements are more diverse than in previous cycles. Improved actuator technology (brushless motors, better force sensing) is producing hardware that can perform dexterous manipulation reliably in unstructured environments. Vision-language-action models — foundation models trained on robotics data rather than text — are solving the generalization problem that made earlier robots brittle. Manufacturing scale is driving hardware costs down curves that make commercial deployment viable in more verticals each year. And venture capital concentration in a small number of robotics companies is producing the organizational capability to commercialize quickly.
The verticals that will reach deployment first are those where the unit economics are clearest and the environment is most structured: warehouse pick-and-pack operations (where Figure AI and Apptronik have commercial deployments), light assembly (where Mercedes-Benz and BMW have been piloting humanoid robots), and logistics sorting (where Amazon is the obvious customer). Elder care and healthcare represent the largest long-term opportunities but face harder technical and regulatory paths.
The platform dynamics are the most interesting long-term question. The companies building robotics operating systems — the frameworks that let developers write applications for different robot hardware — are attempting to create the same value capture point that Apple and Google created with iOS and Android. Whether that model translates to embodied AI is genuinely uncertain.
Conclusion
The robotics transition is real, it's happening faster than most analysts are projecting, and the platform layer — whoever builds the iOS equivalent for embodied AI — will capture disproportionate value as hardware commoditizes.
Tags: #robotics, #automation, #humanoid, #embodied-ai, #manufacturing, #future-of-work