Legacy European chip maker STMicro starts deploying humanoid robots to cut costs

I've written about robots helping build AI Factories that help companies that train AI's that control robots before, and now it looks like robots might be soon helping make chips for robots which will in the end create a virtuous loop. STMicroelectronics is one of the long-standing European chipmakers that does not give up and keeps its fabs open and, more importantly, busy despite all odds. However, high-end pressure from leading chipmakers and emerging Chinese chipmakers on the low-end clearly has its impact, and so the company must find new ways to survive. And, following in the footsteps of companies like BMW, one of them appears to be robots working at fabs, according to Reuters.

STMicroelectronics expects automation to raise efficiency at legacy fabs. At the same time, it plans to retrain workers for more specialized roles. In fact, STMicroelectronics's first humanoid robot designed for fab operations was demonstrated by head of manufacturing Thomas Morgenstern. To add context, the company announced plans to fire some 5,000 workers back in 2024.

In a video shown at the event, a robot was seen loading a silicon wafer carrier into one of the machines involved in a silicon wafer production flow. However, while a humanoid robot loading a wafer is certainly very cinematic, this kind of automation is dramatically below what happens in Class 1 – Class 3 cleanrooms in Arizona, Korea, Texas, and Taiwan, where humans are barely present at all. However, the fabs run by Intel, Samsung, and TSMC were designed this way. What STMicro does is introduce robots to fabs designed decades ago.

According to Morgenstern, this initial robot is only the starting point: over the next few years, STMicroelectronics anticipates deploying more than 100 humanoid robots across its manufacturing facilities. These systems are intended to perform routine and physically demanding activities typically carried out in chip plants, including handling wafers and interacting with processing tools. In modern fabs, this is done by specialized equipment designed to hold, transport, and protect semiconductor wafers, usually in FOUPs (front opening unified pods).

Then again, the initiative targets older semiconductor fabrication plants that are barely competitive against newly built fabs in China in terms of costs, as the latest fabs in the PRC are highly automated. At the same time, they cannot accommodate the latest manufacturing tools to get on par with peers in China. Tearing down such facilities and constructing new ones is also problematic due to high costs, complex regulatory requirements, and negotiations with labor unions in Europe. As a result, automation has turned out to be a practical way to improve productivity without undertaking large-scale rebuilding projects.

Given the current semiconductor industry reality - government subsidized highly-automated fabs in China, highly-competitive fabs in Taiwan, the ongoing onshoring in the US that favors the likes of Texas Instruments, and the growing demand from the defense sector - improving operational efficiency at existing fabs seems to be becoming a strategic priority for European chipmakers. At the end of the day, they can serve demand from both emerging and established players across a variety of industries.

Financial incentives also play a role in shaping these decisions, though. Many older facilities are not eligible for subsidies under the European Chips Act because the plan primarily supports 'first-of-a-kind' projects rather than the modernization of existing production lines.