Google Challenges Nvidia: Launches Its Most Powerful AI Chip Ever

Four Times Faster Than the Previous Generation

Google Challenges Nvidia: Launches Its Most Powerful AI Chip Ever

Google has unveiled its seventh-generation AI processing unit (TPU), codenamed Ironwood, which will be commercially available in the coming weeks after a testing period that began last April.

The company stated that the new Ironwood chip, designed entirely in-house, is capable of handling the most demanding AI tasks, from training large models to running robots and smart assistants in real time, according to a report by CNBC. Google can connect more than 9,200 chips in a single unit (Pod) to accelerate data processing and eliminate bottlenecks in large-scale models.

Direct Competition with Nvidia

With this move, Google aims to break Nvidia's dominance in the AI ​​processor market, where most major companies currently rely on Nvidia's graphics processing units (GPUs). Google is betting that its custom-designed internal architecture will deliver higher performance, greater energy efficiency, and a lower price compared to traditional solutions.

The company explained that the new generation is more than four times faster than its predecessor, while Anthropic, the developer of the cloud model, plans to use up to one million Ironwood chips to power its advanced model. Google announced comprehensive updates to its cloud services to make them faster and more agile, in fierce competition with Amazon Web Services and Microsoft Azure.

Google reported cloud revenue of $15.15 billion in the third quarter of 2025, a 34% year-over-year increase, and raised its capital expenditure ceiling to $93 billion to meet the growing demand for AI infrastructure. Google CEO Sundar Pichai said, "We are seeing tremendous demand for AI infrastructure products, whether they are based on TPUs or GPUs, and we will continue to invest heavily to meet this demand."