Google is expanding its TPU lineup with two new specialized chips targeting confidential model training and increased memory capacity. This move extends Google's hardware portfolio beyond its existing AI accelerators into more diverse deployment scenarios.
Background
Google has been developing custom TPUs since 2016 as part of its strategy to reduce reliance on third-party GPUs for AI workloads. Confidential computing is an emerging priority as enterprises seek to protect sensitive models during training.
- Source
- TechCrunch
- Published
- Nov 10, 2026 at 11:54 PM
- Score
- 6.0 / 10