Google announced two new eighth-generation TPU chips designed to power agentic AI systems. The processors represent the company's latest infrastructure push to support increasingly autonomous AI applications.
Google introduced its eighth-generation tensor processing units (TPUs) with two distinct chip variants optimized for the emerging agentic AI era. The new processors build on Google's history of custom silicon for machine learning workloads.
The chips address different compute requirements within agentic systems, where AI models operate with greater autonomy and decision-making capabilities. Google positions the TPUs as infrastructure built specifically for this shift in AI architecture and behavior.
The announcement reflects broader industry competition in AI hardware acceleration. Companies are increasingly developing custom chips to reduce costs and improve performance for specialized workloads beyond traditional model training and inference.
Google Cloud customers can access the new TPUs through standard cloud services. The infrastructure upgrade aligns with Google's strategy to maintain competitive advantage in AI infrastructure as demand for agentic systems grows across enterprise deployments.
Solar installer Sunrun is piloting a program that pays residential customers hundreds of dollars monthly to use their rooftop solar and battery systems as AI computing infrastructure.
Chipset makers and router manufacturers are preparing Wi-Fi 8, the next wireless standard promising faster speeds and lower latency. Here's what we know about the technology and its timeline.
Intel is investing €5 billion ($5.7 billion) to expand its manufacturing facility in Ireland as the chipmaker races to secure its position in the AI semiconductor market.
Samsung Electronics has moved up the timeline for its first South Korean chipmaking facility in Yongin, targeting operations to begin by 2029 instead of 2030 or 2031.