Google broadens its AI hardware portfolio with two new specialized chips

Google has expanded its custom silicon lineup with two new tensor processing units (TPUs) aimed at specialized AI workloads. The TPU 8tx, developed with AMD, is designed for confidential training and inference, embedding security at the hardware level to protect sensitive data and model weights even during active use. It also features a custom inferencing network that enables on-the-fly model distillation and training, which could help companies train models on proprietary data without moving it 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 so AI models can handle a broader range of information during queries. Both chips are part of Google's eighth-generation TPU family introduced in April. The launch comes as hyperscalers like Microsoft and Amazon, along with AI labs such as OpenAI and Anthropic, develop their own AI-specific semiconductors to reduce reliance on Nvidia, which is facing a vulnerable moment after CEO Jensen Huang admitted to an affair with someone alleged to be a Chinese spy.

(Source：TechCrunch)