Google unveils new AI chips to challenge Nvidia dominance

TechnologyDigital
23 Apr 2026 • 11:23 AM MYT
The Sun Daily
The Sun Daily

For the latest news and features from Malaysia and the rest of the world.

Image from: Google unveils new AI chips to challenge Nvidia dominance

Google’s latest TPU chips, designed for AI training and digital agents, aim to reduce reliance on Nvidia’s dominant GPUs in the competitive semiconductor market.

LAS VEGAS: Google has unveiled its latest generation of custom artificial intelligence chips.

The new Tensor Processing Units (TPUs) are designed to train AI models and power the digital “agents” currently captivating the tech industry.

This move is part of a broader trend where major cloud providers like Google and Amazon are developing their own cutting-edge semiconductors.

They aim to gain control over designs and reduce their reliance on the highly sought-after Nvidia graphics processing units that dominate the AI hardware market.

The high-performance TPUs were a key announcement at Google’s annual cloud computing conference. Google chief executive Sundar Pichai stated that infrastructure must evolve for the era of AI agents.

“In the era of AI agents, infrastructure needs to evolve to take on the most demanding AI workloads,” Pichai said in a blog post. The company is introducing an eighth-generation TPU with a dual-chip architecture.

One new chip is optimised for training the large language models that underpin AI. The other is specifically tailored for the “inference” process, which involves AI reasoning and decision-making.

AI agents are advanced digital assistants capable of independently handling complex computing tasks. Google Cloud unit leader Thomas Kurian confirmed the new TPUs, created in partnership with semiconductor maker Broadcom, will be available later this year.

The announcement intensifies competition in the AI chip sector. Nvidia earlier this year announced new Vera and Rubin GPUs for powering AI, shortly before Amazon unveiled the latest generation of its custom Trainium processors.

Despite developing their own silicon, Google, Amazon and Microsoft continue to integrate Nvidia GPUs into their vast computing infrastructures. This hybrid approach allows them to leverage the best available technology while pursuing greater independence.