As semiconductor manufacturers innovate to support the ever-increasing needs of AI, AI is helping them do so.
Aug. 13, 2025 –
At the recent Leti Innovation Days (LID) 2025 in Grenoble, France, one of the central themes was the paradox of AI and hardware: AI demands new types of chips to scale sustainably, yet AI itself is already playing a critical role in designing and manufacturing those chips. The event showcased the research ambitions of institutions like CEA-Leti and the industrial realities of companies such as GlobalFoundries, which are applying AI not just in laboratories but on the factory floor.
AI’s real-world impact on semiconductor development is tangible at GlobalFoundries’ Dresden facility. In an exclusive interview with EE Times Europe following LID 2025, Frank Jakubowski, senior director for Manufacturing Operations at GlobalFoundries Dresden, detailed how the company is using AI to optimize its complex manufacturing systems.
The Dresden fab, which produces around 880,000 wafers annually across technologies ranging from 22 nm to 55 nm, relies on a 23-kilometer overhead transport system with more than 900 automated vehicles that carry wafers through the different steps in the manufacturing process. “Any vehicle malfunction can stall production tools and create significant capacity loss,” Jakubowski explained. “So we implemented acoustic and visual AI monitoring to evaluate each vehicle in real time, assigning them a health score.”
This predictive maintenance approach allows the system to automatically reroute vehicles to maintenance before they fail. The results are measurable: The fab tripled the mean time between vehicle interruptions and reduced transport-related production losses by 80%.
The AI system also incorporates video analysis to evaluate wheel wear and spacing. Because the vehicles move quickly, still images are insufficient—so the system processes video sequences to flag mechanical degradation early. “The wafers these vehicles carry are destined for applications ranging from automotive to IoT and smartphones,” Jakubowski said. “If one vehicle goes down, it doesn’t just affect throughput—it can disrupt an entire toolchain.”
GlobalFoundries’ efforts are part of a larger push toward fully automated fabs, particularly important in Europe where labor shortages make automation a strategic imperative. The company is now piloting vibration and thermographic sensors to extend predictive analytics beyond the transport system. “We want to use AI to monitor gas delivery lines, chemical supply systems, even exhaust management,” he said. “Any of these could disrupt wafer quality or halt production.”