Google Launches Two Dedicated AI Chips, Expanding Confidential Computing and Memory Capacity

According to TechFlow, on November 10, as reported by TechCrunch, Google launched two eighth-generation Tensor Processing Unit (TPU) variants, TPU 8tx and TPU 8i Max, expanding the application scenarios for its proprietary AI chips.

The TPU 8tx, co-designed with AMD and based on the TPU 8t, targets confidential model training and inference, protecting sensitive data and model weights in use through hardware-level safeguards. Its custom inference network can run inference during training, enabling immediate distillation and training of smaller models.

The TPU 8i Max integrates two inference-focused TPU 8i chips into a single package, doubling memory capacity, processing power, and bandwidth, enabling models to process a broader range of information at query time.

Source: TechFlow