Google launches TPU 8tx and TPU 8i Max chips for secure and high-memory AI tasks
Also covered by 4 sources
What happened
On November 9, 2026, Google introduced two new systems in its eighth-generation TPU lineup: the TPU 8tx and the TPU 8i Max. The TPU 8tx is designed for secure, confidential AI inference and small-scale training, using enhanced security features and lower latency. The TPU 8i Max combines two TPU 8i chips in a single package, doubling compute and memory bandwidth for long-context AI tasks. Both systems are set to be available in 2027 and are aimed at meeting the needs of regulated industries and complex AI workloads.
Key details
- TPU 8i Max integrates two TPU 8i chips, doubling compute, SRAM, and HBM bandwidth within the same rack footprint.
- TPU 8tx uses TPU 8t silicon paired with 5th Gen AMD EPYC hosts and features up to 2x lower latency than TPU 8t.
- Both systems leverage existing eighth-generation TPU silicon with new packaging and system configurations.
- The new TPUs expand Google's lineup to four specialized models for different AI infrastructure needs.
- No pricing, benchmark results, or regional availability details have been published yet.
Why it matters
Organizations in regulated industries and those working with sensitive data can now process confidential workloads with higher security and lower latency using TPU 8tx. AI developers and enterprises running large, memory-intensive models benefit from TPU 8i Max’s increased compute and memory bandwidth without needing more physical space. These changes give Google Cloud customers more specialized options for secure and high-performance AI workloads, with new capabilities expected to be accessible in 2027.