Report: Google launches the strongest AI chip to challenge Nvidia's dominance with custom silicon

Google is launching its most powerful AI chip to date, attempting to capture business from artificial intelligence companies by offering custom silicon solutions. The latest move by the search giant directly targets NVIDIA's dominance in the AI chip market.

On November 6, according to media reports, Google announced that its seventh-generation Tensor Processing Unit (TPU) Ironwood will be available to the public market in the coming weeks, after the chip was launched in April for testing and deployment.

This chip can connect up to 9,216 chips to a single cluster, and Google claims it eliminates "data bottlenecks for the most demanding models," allowing customers to "run and scale the largest and most data-intensive models currently available."

The product marks an important step for Google in the race for AI infrastructure against Microsoft, Amazon, and Meta.

Although most large language models and AI workloads rely on NVIDIA's Graphics Processing Units (GPUs), Google's TPU offers advantages in price, performance, and efficiency as a custom silicon solution. AI startup Anthropic plans to use up to 1 million new TPUs to run its Claude model.

Google's financial report last week showed that its cloud business revenue reached $15.15 billion in the third quarter, a year-on-year increase of 34%, and raised the upper limit of its capital expenditure forecast for this year from $85 billion to $93 billion to meet the surge in demand.

Performance Quadrupled, Major Clients in Line

Reports indicate that the Ironwood chip is entirely developed in-house by Google and is designed to handle a variety of tasks, from training large models to real-time chatbots and AI agents. Google stated that the performance of this chip is more than four times that of the previous generation.

The development of the TPU has been ongoing for ten years. Unlike NVIDIA's GPUs, TPUs belong to the category of custom silicon specifically optimized for AI workloads. Major clients like Anthropic have confirmed their adoption of the chip, highlighting its appeal in high-end AI applications.

Google also launched a series of upgrades aimed at making its cloud services cheaper, faster, and more flexible to compete with larger cloud service providers like Amazon AWS and Microsoft Azure.

Strong Growth in Cloud Business, Record Transaction Scale

Google's cloud business is showing strong growth momentum. The $15.15 billion revenue in the third quarter represents a year-on-year increase of 34%, with growth rates between Azure's 40% and AWS's 20%.

Google revealed that the number of $1 billion-level cloud deals signed in the first nine months of 2025 exceeds the total from the previous two years. This data reflects a surge in demand from enterprise customers for Google's AI infrastructure.

To meet the skyrocketing demand, Google has raised the upper limit of its capital expenditure forecast for this year to $93 billion, an increase of $8 billion from the previous $85 billion.

CEO Sundar Pichai stated during the earnings call:

"We are seeing significant demand for AI infrastructure products, including TPU and GPU-based solutions. This has been one of the key drivers of our growth over the past year, and I believe we will continue to see very strong demand in the future, and we are investing to meet that demand." Reports indicate that this statement highlights Google's long-term commitment to the AI chip business. Against the backdrop of tech giants like Microsoft, Amazon, and Meta heavily investing in AI infrastructure, Google's custom chip strategy provides it with a differentiated advantage in this competition