The company’s seventh-generation TPU chip is custom-built to handle tasks such as large-scale AI model training and high-volume AI inference.