Google Launches Ironwood TPU to Challenge Nvidia in AI Chip War

Google has officially announced the release of its seventh-generation Tensor Processing Unit (TPU), codenamed Ironwood, marking a significant step in its decade-long effort to compete with Nvidia in the AI infrastructure market. The chip, which will be available to the public in the coming weeks, is designed to handle training of large AI models and run real-time AI agents and chatbots. Google claims Ironwood is over four times faster than its predecessor, with each pod capable of connecting up to 9,216 TPUs to eliminate data bottlenecks for demanding models.

The development of TPUs began over ten years ago, initially for internal use before becoming a core part of Google Cloud's public AI infrastructure in 2018. Analyst Stacy Rasgon from Bernstein noted, "Of the ASIC players, Google’s the only one that’s really deployed this stuff in huge volumes." Google does not sell TPUs as physical hardware; instead, customers rent access through Google Cloud, which reported $15.15 billion in cloud revenue last quarter, a 34% year-over-year increase.

Key partnerships underscore the chip's impact: AI startup Anthropic plans to deploy up to 1 million Ironwood TPUs to power its Claude model, supported by a multi-billion-dollar expanded deal with Google. Anthropic’s Chief Product Officer, Mike Krieger, stated, "There is such demand for our models that I think the only way we would have been able to serve as much as we’ve been able to this year is this multi-chip strategy." Google also secured a six-year, $10 billion+ cloud contract with Meta and is involved with OpenAI, though the latter is not using GPUs on Google Cloud.

To meet rising demand, Google raised its 2025 capital expenditure forecast to $93 billion, up from $85 billion, with further increases expected in 2026. CEO Sundar Pichai told investors, "We are seeing substantial demand for our AI infrastructure products, including TPU-based and GPU-based solutions." Additionally, Google revealed Project Suncatcher, aiming to launch solar-powered satellites equipped with TPUs by early 2027 to harness space-based solar energy for AI tasks.

Nvidia, the current market leader, faces growing competition. CEO Jensen Huang initially warned that China could win the AI race due to lower energy costs and regulations but later clarified on social media, emphasizing the need for the U.S. to lead. Analysts from D.A. Davidson highlighted that Google’s TPUs are closing the gap with Nvidia, potentially becoming a direct alternative, while Morgan Stanley pointed to TPU familiarity as a boost for Google Cloud growth.