Google's rolling out most powerful AI chip, taking aim at Nvidia with custom silicon

Google is set to make waves in the tech industry by launching its most advanced chip to date, the Ironwood Tensor Processing Unit (TPU). This announcement comes as the tech giant intensifies its efforts to attract business from artificial intelligence firms with tailored silicon solutions. The Ironwood TPU, which was previously introduced for testing in April, will soon be available for public use, marking a significant step in Google's AI ambitions. Designed for a wide range of applications—from training extensive models to powering real-time chatbots and AI agents—the Ironwood TPU boasts the capability to connect up to 9,216 chips in a single pod. Google claims this configuration effectively removes data bottlenecks for its most demanding models, allowing customers to operate and scale the largest, most data-intensive AI models available today. In a fierce competition for AI infrastructure, Google faces off against major players like Microsoft, Amazon, and Meta. While Nvidia has historically dominated the market with its graphics processing units (GPUs), Google's TPUs represent a shift toward custom silicon that can offer significant benefits in terms of cost, performance, and efficiency. After a decade of development, the Ironwood TPU is reported to be more than four times faster than its predecessor, generating considerable interest from major clients. Notably, AI startup Anthropic plans to deploy up to 1 million of these new TPUs to enhance its Claude model. In addition to the Ironwood chip, Google is also launching a series of upgrades aimed at making its cloud services more competitive against larger players like Amazon Web Services and Microsoft Azure. In its recent earnings report, Google announced a third-quarter cloud revenue of $15.15 billion, which represents a 34% increase from the previous year. In comparison, Azure saw a 40% revenue increase, while Amazon's AWS grew by 20%. To accommodate the heightened demand for its AI infrastructure, Google has raised its capital expenditure forecast for the year to $93 billion, up from $85 billion. CEO Sundar Pichai emphasized the growing need for their AI infrastructure products, stating that both TPU and GPU solutions are crucial drivers of the company's growth over the past year and will continue to be a focus moving forward.