The Coming Build-Out Phase of Physical Autonomy
We are on the cusp of an aggressive "over investment" cycle in robotics and physical AI. In some areas, it has already begun, and society should view this shift with optimism. There is frequent criticism that recent robotics and physical AI funding rounds are "insane". Companies (who are pushing the needle) such as Figure, Apptronik, Physical Intelligence, Skild AI, Field AI, and others are raising at unprecedented scale. Given the magnitude of this early capital allocation, it is understandable why some people hesitate. This period is required because it enables the early construction of a future where autonomous machines expand what humanity can build, repair, and create.
When societies shift into each new technological era, the early investment cycle always looks disproportionate because the underlying infrastructure must be built before the gains appear. Rail networks, electrical grids, the early semiconductor fabs, and the first generation of internet data centers all seemed oversized in their time, yet each clearly became the backbone for entire eras of growth. The current wave of investment into artificial intelligence follows a similar pattern. Robotics and physical autonomy follow the same logic.
Companies will rise and fall, capital expands and evaporates, and the short-term scoreboard often looks irrational. Beneath that churn, real assets accumulate such as infrastructure, supply chains, trained talent, and operational expertise. These become the building blocks that enable the next generation of companies to scale far faster than the first. What looks like excess in the moment will become the groundwork for an entirely new industrial base.
In the near term, the important technical work will be in deployment. Robotics needs factory ready orchestration platforms, reliable edge inference, and rollout patterns that let teams integrate systems into active operations with controlled downtime and predictable throughput. Generalized foundation models for robotics will continue to be area of advancement. Significant progress is underway here already, as seen in efforts from Nvidia GR00T, Skild AI, and Physical Intelligence. A parallel trend is the rapid maturation of next generation robotic hardware, including early general purpose mobile platforms, that can support broader task coverage and tighter integration with foundation model based control systems. The ongoing reshoring of manufacturing in the United States also creates a clear opportunity, but only if robotic systems integrate cleanly with modern industrial software stacks and meet the reliability standards required in production environments.
The upside is safer and more resilient industries, abundant production capacity, accelerated scientific progress, stronger healthcare and elder care systems, and a physical world that adapts to human needs rather than the other way around. I am deeply optimistic about this future, and believe that this next investment cycle into physical AI and robotics is required for humanity to unlock the progress that lies ahead.