Google Expands AI Hardware Portfolio with Two Specialized TPU Chips

Google is stepping up its chipmaking efforts as it seeks to reduce its dependence on third-party silicon for its cloud computing business. On Monday, the company unveiled two new variants of its tensor processing units (TPUs)—in-house chips designed for AI workloads—broadening its portfolio into additional use cases: confidential inference and greater memory capacity for large workloads.

The new units, the TPU 8tx and TPU 8i Max, are part of the company's eighth-generation chip family introduced in April 2026. They build on the capabilities of that generation, which emphasized greater inference capacity and support for large-scale training requirements.

TPU 8tx: Confidential Training and Inference

The TPU 8tx, designed in collaboration with AMD, is based on the TPU 8t chip and is purpose-built for confidentially training and running AI models, 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.

The chip also features a custom inference network that allows it to run reasoning alongside training, enabling it to distill and train smaller models on the fly. According to Google, this capability could prove useful in scenarios where an organization wants to train a model on proprietary data without moving that data between devices.

TPU 8i Max: Doubling Down on Inference

Meanwhile, the TPU 8i Max essentially combines two of the inference-focused TPU 8i chips the company launched in April. It doubles the memory capacity, processing power, and bandwidth—all within a single package. The result is a chip that allows AI models to handle a broader range of information during queries.

A Competitive Landscape Heating Up

The launch comes as hyperscalers such as Microsoft and Amazon, along with AI labs like OpenAI and Anthropic, are developing their own lines of AI-specific semiconductors and accompanying software to reduce their reliance on Nvidia's market dominance.

The news also arrives at a particularly vulnerable moment for Nvidia. Over the weekend, CEO Jensen Huang admitted to having an affair with someone alleged to be a Chinese spy.

via TechCrunch AI