Physical AI is a new frontier where physical devices meet artificial intelligence, promising significant commercial applications beyond the scope of traditional humanoid robots. This technology enables robots to combine autonomy with hardware that manipulates objects in the real world, utilizing sensors to perceive their surroundings. Unlike older, deterministic robots that perform repetitive, precise tasks, physical AI robots can complete varied, complex tasks and adapt to changing circumstances.
Over 4.7 million industrial robots were operational in 2024, with this number increasing by over 500,000 annually, double the rate of a decade ago. These AI-powered robots are shifting from "high-volume, low-variation" environments, like automotive production lines, to "high-variation, low-volume" settings, making them viable for dynamic and small-scale operations. Siemens' global head of manufacturing, Stephan Schlauss, noted that AI-enabled robots on assembly lines reduce automation costs by 90% and enhance productivity by empowering manual workers with AI-guided systems.
Robots learn through various methods, including observing human demonstrations, watching videos, and trial-and-error in real or simulated environments. While one-shot learning algorithms, which require only a single demonstration, are still emerging and complex to design, current training can still be extensive, with hundreds of hours of demonstrations sometimes needed. Amazon, for example, uses over 1 million robots in its fulfillment centers, with AI capabilities providing flexibility. Its Vulcan robot, with feedback sensors, can pick and stow three-quarters of Amazon's items and continuously learns how items behave when handled.
The sector is growing rapidly, with 381 deals transacted in the first quarter of 2025, a 20% increase from the same period in 2024. The disclosed value for these transactions was $21.6 billion, though this represents only 15% of total deals due to the private nature of many transactions. Softbank's $5.4 billion acquisition of ABB's robotics arm illustrates increased industry focus on intelligent robots. Key challenges include the "data dilemma," where good data is essential but often costly and difficult to collect, especially in specialized fields like surgical procedures. Training also remains a slow and expensive process.
Despite these challenges, the advantages of physical AI are clear: reducing danger in physical jobs, improving efficiency, and filling workforce gaps. The World Economic Forum emphasizes that companies must redesign workflows to maximize the benefits of these intuitive, AI-enabled robots. As AI advances, factory and warehouse staff will transition from manual tasks to working alongside robots, managing them and performing simple maintenance. These robots are also becoming more accessible due to increased processing power and less reliance on fixed hardware, bringing cutting-edge technologies within reach for companies with smaller budgets.
