Physical AI and the new energy calculus in manufacturing

Physical AI can make industrial electrification and automation more competitive by improving both the intelligence and energy efficiency of factory systems.

• Industry consumes nearly 40% of final global energy demand, while global electrification has stagnated at around 20%, leaving roughly 80% of demand dependent on fossil fuels; industry also represents around two-thirds of energy-demand growth.
• Physical AI is advancing from deployed applications such as predictive maintenance and industrial sensing, through pilots in digital twins and process monitoring, toward autonomous multi-robot and human collaboration; AI-enabled products already represent a mid-single-digit to low-teens share of leading factory-automation revenues.
• Robeco’s conversations with industrial technology companies identify Siemens, Schneider Electric, Hitachi and Nexans as examples using robots, digital twins, AI agents and sensor networks to improve logistics, equipment diagnosis, power distribution, failure prediction and grid optimisation.
• Robot actuators are a major but underappreciated energy constraint: motor efficiency of about 80% can fall to approximately 40% after gearboxes and transmissions, across a global installed base of around 4.3 million industrial robots.
• AI-optimised motion planning reduced energy consumption by 74.8% in Fanuc pick-and-place operations, while humanoid robots from Tesla, Agility Robotics and Apptronik at Mercedes-Benz currently run only two to four hours per charge; newer actuators may deliver three to five times higher torque-to-weight ratios and push efficiency above 90%.
• Physical AI therefore creates investment demand for power-efficient semiconductors, control software, actuators, motors, smart grids, storage and power electronics, with incumbents that control real-world operational data potentially gaining an advantage.

Takeaway: Investors should assess physical-AI companies not only by their software and automation capabilities but also by how efficiently they convert electricity into useful factory output; the opportunity is significant, but the referenced companies are illustrative rather than recommendations or confirmed holdings.