Feed
Google Unveils Two Specialized TPUs for Confidential AI Training and Memory
AI

Google Unveils Two Specialized TPUs for Confidential AI Training and Memory

10 Nov 2026, 9:24 pm · 10d ago · 1 min read · TechCrunch AI

Google has expanded its custom AI hardware lineup by introducing two new specialized Tensor Processing Units. These next-generation accelerators are specifically designed to address targeted enterprise workloads, focusing on confidential AI model training and significantly higher memory capacity. The move allows cloud customers to train sensitive machine learning models in secure environments while handling memory-intensive artificial intelligence tasks more efficiently. By diversifying its custom silicon portfolio, Google aims to reduce reliance on third-party GPU providers like Nvidia and strengthen its Google Cloud infrastructure for large-scale enterprise AI deployments across global markets.