Our AI method has accelerated and optimized chip design, and its superhuman chip layouts are used in hardware around the world. AlphaChip was one of the first reinforcement learning approaches used to solve a real-world engineering problem. It generates superhuman or comparable chip layouts in hour…