Google Cloud Launches Arm-Based CPUs to Power AI Workloads

Google unveiled its first Arm-based CPU design to power generative AI workloads in data centers at its Google Cloud Next 2024 event this week in Las Vegas.

Named Axion, the new CPU line is designed to offer better performance while also being energy efficient. 

tThe new Axion chips offer up to 50% better performance than current-generation x86-based chips, according to Google.

The company also claims the new CPUs are 60% more energy efficient than rival hardware solutions.

“Axion processors combine Google’s silicon expertise with Arm’s highest performing CPU cores to deliver instances with up to 30% better performance than the fastest general-purpose Arm-based instances available in the cloud today,” Amin Vahdat, Google’s vice president and general manager for machine learning, systems and cloud AI wrote in a blog post.

Businesses won’t be able to buy the chips. Instead, Axion CPUs will be made available to customers to rent via Google Cloud services including Google Kubernetes Engine and Dataflow later this year.

Google is already using current generation Arm-based chips to power services including the Google Earth Engine, the YouTube Ads platform and BigQuery, Google Cloud’s data warehouse platform, with plans to deploy them on Axion soon.

Related:Google Launches Gemini Next-Generation Large Language Model

Google Cloud has launched a preview of Arm-based instance support in its Migrate to Virtual Machines service, allowing customers to migrate existing applications and workloads to virtual machines on its cloud platform.

Google follows Microsoft and Amazon in launching its own Arm-based CPUs. Last November, Microsoft unveiled Cobalt designed for general-purpose cloud workloads while Amazon revealed Graviton4, the latest in its line of custom Arm-based chips.

“Google’s announcement of the new Axion CPU marks a significant milestone in delivering custom silicon that is optimized for Google’s infrastructure and built on our high-performance Arm Neoverse V2 platform,” said Rene Haas, Arm’s CEO.

Google Cloud also unveiled updates to its Tensor Processing Units (TPUs), custom hardware units designed to power enterprise-accessible hypercomputers, including making its latest version generally available. 

“We’re thrilled to announce the general availability of Cloud TPU v5p, our most powerful and scalable TPU to date,” said Mark Lohmeyer, Google Cloud’s vice president and general manager for compute and AI Infrastructure. “TPU v5p is a next-generation accelerator that is purpose-built to train some of the largest and most demanding generative AI models."
Google Cloud first unveiled the Cloud TPU v5p last December. The hardware was used to power training for Google’s Gemini flagship large language model.

Related:Google researchers create an AI system to speed up chip design