Google broadens its AI hardware portfolio with two new specialized chips

Google announced two new tensor processing units (TPUs) on Monday, expanding its in-house AI chip portfolio. The new TPU 8tx and TPU 8i Max chips are designed to enhance confidential model training and handle larger workloads, respectively. These chips are part of Google's eighth generation of TPUs, first introduced in April 2026.

What the new TPUs offer

The TPU 8tx, developed in collaboration with AMD, is specifically designed for confidential training and running AI models. It integrates security directly into the hardware layer, ensuring that sensitive data and model weights remain protected even during active use. This chip also features a custom inferencing network that allows it to perform reasoning alongside training, enabling on-the-fly distillation and training of smaller models. This capability is particularly useful for companies that want to train models on proprietary data without transferring the data between devices.

The TPU 8i Max, on the other hand, combines two inference-focused TPU 8i chips into a single package. This combination doubles the memory capacity, processing power, and bandwidth, allowing AI models to handle a larger range of information during queries.

The broader industry context

The introduction of these new TPUs is part of Google's broader strategy to reduce its dependence on third-party chips for its cloud computing business. By developing its own AI-specific semiconductors, Google can optimize its hardware for specific AI workloads. This move also aligns with a broader industry trend, as other hyperscalers like Microsoft, Amazon, and AI labs such as OpenAI and Anthropic are also developing their own AI-specific chips to reduce reliance on Nvidia.

Google's TPUs have been a significant part of its AI infrastructure since their introduction in 2016. The eighth generation, introduced in April 2026, marked a significant leap in inference capacity and support for large training requirements. The new TPU 8tx and TPU 8i Max chips build on this foundation.

Why these chips matter for everyday users

For everyday users, these advancements mean more secure and efficient AI services. For example, the TPU 8tx's ability to train models on proprietary data without transferring it could lead to more personalized and secure AI applications, such as customized healthcare or financial services. The TPU 8i Max's increased capacity could enable more complex AI models to run efficiently, potentially improving services like real-time language translation or advanced data analytics.

In the healthcare sector, for instance, the TPU 8tx could be used to train models on patient data without compromising privacy, leading to more accurate and personalized medical diagnoses. In the financial sector, the TPU 8i Max could enable more sophisticated fraud detection systems by processing larger datasets more efficiently.

What you can do today

While these new chips are primarily targeted at businesses and developers, their impact will eventually trickle down to everyday users. If you use Google Cloud services, keep an eye out for announcements about new AI features powered by these chips. For now, you can explore Google's existing AI services, such as Google Cloud's AI Platform, to see how they might benefit from these hardware advancements in the future. If you're a developer, consider experimenting with Google's TPUs to see how they can enhance your AI projects.

For those interested in the technical aspects, Google's documentation and developer resources provide detailed information on how to integrate TPUs into your workflows. Additionally, Google Cloud offers a range of AI and machine learning services that can be enhanced by these new chips.