Google posted at 13:43 UTC on 9 November 2026, which is 21:43 in Beijing the same day. The post says the eighth-generation TPU line is expanding by two systems built for the agentic era, bringing the count to four distinct eighth-generation TPU systems launched in a single year.
[1]
The post assigns jobs to the two new systems. TPU 8i Max was built for massive context windows and the highest compute density per rack. TPU 8tx is optimized for secure and confidential inference. The post says the four systems can still be used across workloads, and that they give a purpose-built accelerator for every stage of the agentic lifecycle. The explanation is left to Amin Vahdat, senior vice president and chief technologist for AI and infrastructure, in the video attached to the post.
[1]The post does not give a price, a ship date, or a list of customers already using the systems, and it does not give measured comparisons with the earlier eighth-generation systems. It does not name the other two of the four. The video frame shows Amin Vahdat holding a chip, and the post text does not name that chip.
[1]要点
- The post counts four distinct eighth-generation TPU systems in one year, two of them new
- 8i Max is for long context and rack density, and 8tx is for confidential inference
- Price, ship date, and the names of the other two systems are absent