Google TPU 8tx and 8i Max add confidential computing and massive context to the eighth-generation lineup

Google expanded its eighth-generation TPU portfolio on November 9, 2026, with two new systems announced by Amin Vahdat, its SVP and chief technologist for AI and infrastructure. TPU 8tx is built for secure, confidential inference and small-scale training: it uses the same silicon as the training-focused TPU 8t, paired with 5th Generation AMD EPYC CPU hosts, so sensitive data and model weights stay protected and isolated even while in active use. TPU 8i Max attacks the memory wall instead, combining two TPU 8i chips on a single package for double the compute, double the SRAM and double the HBM bandwidth of TPU 8i. The additions make 2026 the first year Google has introduced four TPU systems, and Google says TPUs now serve nearly 100 customers.

The hardware arrives with an openness push. TorchTPU was open-sourced last month so PyTorch code and Hugging Face models run natively on TPUs; a TPU Agent Suite, now in preview, plugs into coding tools such as Claude Code and Cursor; the XLA Serving Engine is open source; and TPU Sync, also open-sourced, cuts KV-cache transfer latency and time to first token by more than 90%, Google says. Cloud Pathways is now generally available on Google Kubernetes Engine. Google published no prices or full specifications for the new systems; Bloomberg reported Vahdat saying both are expected in 2027.