Google Cloud Unveils New AI Chips to Challenge Nvidia

Google has introduced its latest Tensor Processing Units, which are engineered to be faster and more economical than their predecessors. The company is positioning these new chips as a direct competitor to Nvidia's offerings in the artificial intelligence hardware market.

Despite the launch of its own improved silicon, Google Cloud is maintaining its partnership with Nvidia. The company will continue to provide access to Nvidia's hardware through its cloud platform. This dual strategy suggests Google is not yet ready to fully commit to its proprietary chips for all AI workloads.

The decision to embrace both its own TPUs and Nvidia's products reflects the current market dynamics. While Google aims to build a competitive advantage with its custom hardware, it also recognises the widespread industry reliance on Nvidia's technology. For now, customers will have a choice of accelerators within the Google Cloud ecosystem.

• ·Google has launched new Tensor Processing Units (TPUs) that are faster and cheaper than previous versions.
• ·The new chips are designed to compete directly with Nvidia's hardware in the AI market.
• ·Google Cloud will continue to offer Nvidia's chips alongside its own proprietary hardware.

Al Coxen covers AI hardware, inference infrastructure, and the companies building the compute layer powering modern AI for the LiberaGPT team. With a decade reporting on semiconductors and cloud, he focuses on the physical reality behind the intelligence revolution.