The BareStory
Google announced on Thursday that it will make its seventh-generation custom artificial intelligence chip, the Ironwood Tensor Processing Unit (TPU), widely available in the coming weeks. The company stated that the new chip is more than four times faster than its predecessor and is designed for intensive AI workloads, such as training large models and operating real-time AI agents.
The release is part of the company's effort to compete in the AI infrastructure market against rivals like Amazon, Microsoft, and the prevalent chip supplier Nvidia. AI startup Anthropic intends to use up to one million of the new Ironwood chips. Anthropic’s Chief Product Officer, Mike Krieger, stated that a multi-chip strategy involving Google, Amazon, and Nvidia was necessary to meet demand.
The move comes as Google experiences growth in its cloud division. The company reported its third-quarter cloud revenue was $15.15 billion, a 34% increase from the prior year. In response to demand, Google increased its 2025 capital expenditure forecast to a high of $93 billion. CEO Sundar Pichai said that substantial demand for the company's AI infrastructure products is a key driver of growth.
Google has been developing TPUs for over a decade and offers them as a cloud service rather than selling the hardware directly. According to an analyst from Bernstein, Google is the only major cloud provider to have deployed its own custom AI chips in such large volumes. The company also remains a major buyer of Nvidia's GPUs. On Tuesday, Google also announced Project Suncatcher, an initiative to explore using solar-powered satellites equipped with TPUs.
How it may affect me
As a U.S. reader:
• The development of faster AI chips could lead to improved performance and new capabilities in consumer-facing applications and services powered by artificial intelligence.
• Increased investment and competition in the AI hardware market among major tech firms may impact the technology sector and related financial markets.
• Competition among Google, Nvidia, and others for AI customers may influence the long-term cost and variety of AI-driven products available to the public.