 |
| Illustration: The Verge |
Google is stepping up its game in the tech industry by developing a custom Arm-based CPU named Axion, specifically designed to bolster its AI initiatives within data centers. This move is complemented by the launch of a new and more potent iteration of its Tensor Processing Units (TPU) AI chips. The introduction of the Axion CPU marks a significant step for Google as it aims to enhance the performance of its AI workloads, with plans to make these chips available to Google Cloud's business clients in the near future.
The tech giant has already begun utilizing the Axion chips across various services, including powering YouTube advertisements, the Google Earth Engine, and more. Mark Lohmeyer, the Vice President and General Manager of Compute and Machine Learning Infrastructure at Google Cloud, emphasized the ease with which customers can transition their existing workloads to the Arm platform, thanks to Axion's design based on open standards.
Google's new CPU is set to be integrated into several cloud services, such as the Google Compute Engine, Google Kubernetes Engine, and others, promising a performance boost of 30% over traditional Arm chips and 50% over current Intel processors.
In addition to the CPU developments, Google is also enhancing its TPU AI chips, aimed at offering an alternative to Nvidia's GPUs for AI acceleration. The latest version, TPU v5p, is designed to handle the training of large and complex generative AI models, boasting over 8,960 chips in a single pod, more than double the previous version.
This announcement follows similar moves by Microsoft and Amazon, who have both introduced custom silicon chips for their cloud infrastructures, signaling a broader trend towards self-reliance and innovation in the tech industry. Google's approach, however, focuses on providing these chips through its cloud services rather than direct sales, highlighting its strategic emphasis on cloud computing and AI technologies.
With these advancements, Google positions itself alongside its competitors, reducing dependency on external suppliers like Intel and Nvidia, and underscoring its commitment to leading in both hardware innovation and cloud services.