Frontier Flux Original
Google adds TPU 8tx and TPU 8i Max, its third and fourth eighth-generation TPU systems
Google announced TPU 8tx for confidential inference and TPU 8i Max, a dual-chip package for very large context windows, on 9 November 2026. TechTarget reports no dates or pricing yet, and the performance figures are Google's own.

Frontier Flux / AI-generated
Google on 9 November 2026 added two systems to its eighth-generation TPU lineup, according to a Google blog post by Amin Vahdat, SVP and Chief Technologist for AI and Infrastructure. TPU 8tx targets secure and confidential inference, and TPU 8i Max targets very large context windows and rack density. With TPU 8t and TPU 8i from April, Google says this is the first time it has introduced four TPU systems in one year.
TPU 8tx uses the same silicon as TPU 8t, paired with 5th Generation AMD EPYC CPU hosts, and keeps data and model weights protected in hardware while they are being processed. Google claims up to 2x lower latency than TPU 8t with a near-zero performance penalty, and says teams can fine-tune, distill and run reinforcement learning on the same confidential system. TPU 8i Max puts two TPU 8i chips on one package for double the compute, SRAM and HBM bandwidth of TPU 8i in the same rack footprint. TechTarget reports 768 MB of on-chip SRAM and 17.2 TB/s of HBM bandwidth, and notes Google gave no availability dates or pricing for either system. The Verge confirmed the two new versions and their intended customers.
Key facts
- Primary: Google blog post, 9 November 2026
- Independent: TechTarget and The Verge
- TPU 8tx: confidential inference and small-scale training for regulated firms and governments
- TPU 8i Max: two TPU 8i chips on one package, aimed at million-token contexts
- Software: open-source TPU Sync model-state layer, Agent Compute Engine for idle agent sessions, and Cloud Pathways and GKE dynamic slicing now generally available
- Customers: Google says TPUs have nearly 100 customers
- Not announced: availability dates and pricing for the two new systems
How to read it
These are new system configurations built around existing eighth-generation chips, not a new chip design. The latency, density and benchmark claims come from Google and have not been tested by outside labs. TechTarget reports that TPU 8t and 8i are still due to come online by the end of this year, so the new variants arrive later than that.
Independence
Frontier Flux is not affiliated with Google, AMD, TechTarget or The Verge. Performance figures are Google's claims and have not been independently verified by Frontier Flux.

