The key facts
- The TPU 8tx and TPU 8i Max are specialized versions of existing eighth-generation chips, not new designs
- The TPU 8tx targets government agencies and regulated industries, while the TPU 8i Max is aimed at tasks with large contexts
- Google Cloud revenue in the third quarter grew 103% to $30.7 billion; Alphabet shares rose more than 3%
Two new configurations based on the eighth generation
On Monday, Google unveiled two new configurations of its own AI processors — the TPU 8tx and the TPU 8i Max. According to Bloomberg , citing the company's announcement, these are specialized versions of already existing eighth-generation chips (tensor processing units), not entirely new designs. Bloomberg was the first to report the launch of the systems, after which Google published the official details.
TPU 8tx: data protection during processing
The TPU 8tx is intended for government agencies and companies in regulated industries that need to protect confidential information directly during processing. The system is built on the same silicon as the TPU 8t and is complemented by customized AMD processors; hardware mechanisms isolate data and AI model parameters while they are actively in use. According to Google, on some operations the system delivers up to a twofold speedup compared with the TPU 8t with almost no performance loss from the security measures.
TPU 8i Max: large contexts and doubled power
The TPU 8i Max is designed for AI systems that handle large context windows — for example, entire codebases or hours of video. Google combined two TPU 8i processors in a single package, doubling the compute power, on-chip memory, and high-bandwidth memory capacity. The configuration is aimed at demanding tasks: coding assistants, mathematical reasoning, scientific research, and drug discovery.
Expanding the lineup and the business context
With the two new products, Google's eighth-generation chip portfolio has grown to four products: the TPU 8t targets large-scale model training, while the standard TPU 8i is aimed at reasoning and real-time inference. The company originally developed TPUs for its own products — Search, YouTube, and the Gemini models — and later opened access to them through Google Cloud and began shipping TPU systems directly to customers' data centers. After the announcement, Alphabet shares rose more than 3% in early US trading.
Competition with Nvidia and cloud business figures
The expansion of the lineup strengthens Google's challenge to Nvidia's dominance in the AI chip market. The chip business is becoming an increasingly significant part of the cloud business: Google Cloud revenue in the third quarter grew 103% year over year to $30.7 billion, and Alphabet named sales of TPU systems a significant factor in that growth. The company expects continued growth in TPU sales in the fourth quarter and a more substantial revenue increase in 2027.

