Google fires shots in the AI chip "ground war": Recruiting new cloud providers to reduce reliance on NVIDIA

Odaily Planet Daily News Google is mounting a competitive offensive against NVIDIA in the AI chip market, attempting to expand the commercial application of its self-developed Tensor Processing Units (TPUs) to challenge NVIDIA's long-standing dominance in the AI computing market.

Google is currently both a significant customer of NVIDIA's AI server chips and is also developing its own AI chip business to compete with and potentially replace NVIDIA's GPUs. Previously, almost all Google TPUs were deployed in its own data centers, with usage available to customers only through Google Cloud rental services.

However, starting this year, Google has changed its strategy, planning to sell TPU chips to emerging cloud service providers (Neoclouds). These new cloud service companies primarily focus on providing AI computing resources and represent a key customer segment driving the demand growth for NVIDIA GPUs.

Analysts believe that this move by Google signifies TPU's transition from internal specialized infrastructure to a broader commercial chip platform. By enticing Neoclouds to adopt TPUs, Google aims to reduce AI companies' reliance on the NVIDIA GPU ecosystem and expand its own influence in the AI infrastructure field.

NVIDIA currently dominates the AI chip market with its CUDA software ecosystem, high-performance GPUs, and vast developer network. As major tech companies like Google, Amazon, and Microsoft continue to advance their self-developed AI chips, competition in the AI computing market is shifting from a single GPU supply model to a diversified chip ecosystem battle. (The Information)