AES says its Maximo system has crossed a useful threshold for real-world construction robotics: 100 megawatts of solar panel installation at the Bellefield 1 project in California. According to the company, that work was completed by a fleet of four robots operating in parallel, moving the system from pilot-style validation into sustained commercial use.

The milestone matters because physical AI demos often stall before they reach jobsite scale. AES is framing Maximo differently. On its product page, the company says the robot can install panels in half the time and at half the cost of standard processes, while also reducing the heavy repetitive lifting that drives construction injuries. It also says Maximo is being deployed across additional U.S. projects.

NVIDIA highlighted the project during National Robotics Week as an example of industrial robots moving from simulation into field deployment. The company said Maximo was developed with accelerated computing, Omniverse libraries, and Isaac Sim, giving AES a software stack for planning, training, and deployment rather than a one-off hardware demo.

There are still open questions around how broadly the economics translate beyond AES' own pipeline. But a 100 MW installation is a more concrete benchmark than most robotics announcements. For physical AI, the signal here is less about humanoids and more about specialized machines quietly taking on repetitive, dangerous work at utility scale.