← Back to GPN · All articles

Google Adds Two New Eighth-Generation TPUs for Secure and High-Context AI

OFFICIAL Written by GPN · Nov 9, 2026, 3:00 PM ET · · Tech

Google has added two more systems to its eighth-generation TPU line-up: TPU 8tx and TPU 8i Max. It is the first time the company has introduced four TPU systems in a single year, following the TPU 8t for large-scale training and TPU 8i for real-time reasoning announced at Google Cloud Next in April.

TPU 8tx is built for secure and confidential inference, aimed at regulated enterprises and governments running frontier or proprietary models alongside sensitive data. Built on the same silicon as TPU 8t and paired with 5th Generation AMD EPYC CPU hosts, it builds security into the hardware layer so data and model weights stay protected and isolated even while in active use, and it delivers up to 2x lower latency than TPU 8t with what Google calls a near-zero performance penalty. Teams can also fine-tune models on private data, distill large models into smaller ones and train agents through reinforcement learning on the same system.

TPU 8i Max takes the opposite approach, attacking the memory wall. Google combined two low-latency TPU 8i chips into a single package, doubling the compute, SRAM and HBM bandwidth of TPU 8i so large workloads — dozens of agents reasoning across a codebase or hours of video at once — do not stall. Because it fits twice the silicon in the existing rack footprint, data centres get more computing power per square foot and per watt.

Alongside the hardware, Google is opening up TPU development: it open-sourced TorchTPU so PyTorch code and Hugging Face models run natively on TPUs, put a new TPU agent suite into preview, and open-sourced the XLA Serving Engine and TPU Sync, a model-state layer for TPU clusters. Cloud Pathways is now generally available on Google Kubernetes Engine, and GKE dynamic slicing is generally available too.

Facts gathered from

• Google

More on this

Share