Google has announced its most advanced artificial intelligence chip to date, marking its biggest step yet in the race for global AI infrastructure. The company on Thursday unveiled the seventh generation of its tensor processing unit (TPU), called Ironwood, which is designed to handle the most demanding AI workloads and compete directly with Nvidia’s industry-leading GPUs.

After being introduced for testing and limited deployment earlier this year, Ironwood TPU will be made widely available for public use in the coming weeks. Built entirely in-house, the new chip is capable of managing both the training of large models and the operation of real-time AI tools such as chatbots and agents.

According to Google, the Ironwood architecture combines 9,216 chips in a single pod, effectively eliminating data bottlenecks and allowing customers to run and scale large-scale, data-intensive AI models. The company claims that Ironwood performs up to four times faster than the previous TPU generation, offering big improvements in speed and energy efficiency.

AI startup Anthropic, one of Google’s key partners, is one of the first major customers of the new chips. Google said Anthropic plans to use one million Ironwood TPUs to run its cloud AI models, indicating strong demand for the new hardware among top AI developers.

The launch comes as Google is ramping up efforts to grow its cloud computing division and establish itself as a leading AI infrastructure provider. Along with the Ironwood rollout, Google is introducing a series of upgrades aimed at making its cloud services faster, cheaper and more flexible as it competes with Microsoft Azure and Amazon Web Services (AWS) for enterprise customers.

Google Cloud is seeing strong growth, with the company reporting revenue of $15.15 billion in Q3 2025, up 34 percent year-over-year. During the same period, Microsoft Azure recorded 40 percent growth, while Amazon’s AWS grew 20 percent.

To keep pace with growing AI demand, Google has also increased its capital expenditure forecast for 2025 to $93 billion, up from the previously announced $85 billion.

“We are seeing substantial demand for our AI infrastructure products, including TPU-based and GPU-based solutions,” Google CEO Sundar Pichai said during the company’s recent earnings call. “This has been one of the key drivers of our growth over the past year, and we continue to see very strong demand going forward.”

Google’s TPUs have been in development for nearly a decade as part of its long-term strategy to create custom silicon optimized for AI workloads. While Nvidia’s GPUs continue to dominate the market, Google’s latest innovation shows a strong push to offer an alternative that promises better scalability, efficiency, and cost benefits for enterprises and AI developers.