Google unveiled two new tensor processing units (TPUs) on Monday: the TPU 8tx, designed with AMD for confidential training and inference, and the TPU 8i Max, which puts two inference chips in one package for workloads that need more memory. Both belong to the eighth generation of TPUs that Google introduced in April, which brought more inference capacity and support for large training jobs.
The TPU 8tx is based on the TPU 8t chip and is meant for training and running AI models confidentially, Google said in a blog post. The chip “builds security directly into the hardware layer, keeping sensitive data and model weights protected and isolated even while in active use,” the company wrote.
It also has a custom inferencing network that lets it run reasoning alongside training, so it can distill and train small models on the fly. Google said that could help a company that wants to train a model on its proprietary data without moving the data between devices.
The TPU 8i Max combines two of the inference-focused TPU 8i chips launched in April into a single package, doubling memory capacity, processing power and bandwidth. According to TechCrunch, the result lets AI models handle a larger range of information during queries.
Google builds its own chips so its cloud computing business depends less on chips made by other companies. Microsoft and Amazon are developing their own AI-specific semiconductors and accompanying software, as are the AI labs OpenAI and Anthropic, all to reduce their reliance on Nvidia, TechCrunch reported.
Related InsightTrack Analysis
- The Sovereignty Gap: Why Canada’s Agentic AI Boom Depends on Foreign Model Providers
- Canada’s Sovereign AI Push Meets the Chip Wall
- Agent Orchestration Is Reshaping How AI Actually Gets Work Done
Source
Google broadens its AI hardware portfolio with two new specialized chips | TechCrunch
