Physical AI Powering the New Age of Industrial Operations 2025

Expanding the boundaries of automation (illustrative) FIGURE 1 Process characteristicsPredictable processes and known environmentUnpredictable processes and known environmentUnpredictable processes and new environmentAutomation potential(Physical) AI considerably increases the automation scope in industrial operations Training-based robotics (e.g. simulation-based AI training)Rule-based robotics (e.g. AI-supported coding) Context-based robotics (e.g. zero-shot learning) Source: BCG, World Economic Forum, expert interviews. How to interpret this chart The area of the chart maps out physical tasks (e.g. assembly steps, material handling, packaging) within a factory or warehouse. These tasks are categorized along two dimensions: Automation potential (y-axis) is indicated through colour shading: Process characteristics (x-axis) are defined by parameters such as object position, orientation and size, and if the system operates in a known or new environment. Illustrative target state along different process characteristics: → Predictable processes: Parameters are either constant or vary only within a tightly controlled range – enabling deterministic, repeatable execution without the need for adaptive behaviour. → Unpredictable processes: Parameters vary significantly or cannot be anticipated. → New environments: Scenarios, layouts, objects or tasks outside the robot’s training distribution (e.g. a different factory line, unfamiliar parts or altered warehouse layout).Grey: Tasks already automatable with today’s rule-based robotics Blue: Additional scope unlocked by physical AI Navy: Illustrative share expected to remain manual in the near term 8 Physical AI: Powering the New Age of Industrial Operations