That headline isn’t great, chip manufacturing is already extremely automated. Wafers can flow through the whole manufacturing process without human intervention (as long as everything goes right). Wafer handling within tools is handled by robots, and wafers are transported through many fabs by robots that run overhead. Automated systems handle the scheduling of product lots to tools and so on. If I am physically handling a wafer something went very wrong…

However, there is definitely still plenty of room for additional automation. Engineers will come up with plenty of ways to use the computational capability of those GPUs. As an example, some process control is still pretty manual. People manually review inline metrology data and make adjustments off of it for some processes. Automatic process control systems exist, but aren’t used for everything. The level of risk and complexity involved in process adjustment can make implementing a dumb process control model rather difficult. There is potential for AI/machine learning to expand automation in process control. I’m sure there is also plenty of room to improve the systems surrounding things like lot scheduling to tools that could improve overall efficiency of the fab, plus plenty of other things I haven’t thought of here.