Microsoft partners with Arm for in-house cloud CPUs

Tech giant Microsoft has unveiled a new in-house chip based on Arm architecture to run general-purpose compute workloads in the cloud.

Microsoft said it will roll out the Azure Cobalt 100 CPU next year into its data centres to power services such as Microsoft Copilot or its Azure OpenAI service.

In addition to the chip based on Arm architecture, Microsoft announced the Azure Maia AI Accelerator, designed for generative AI applications.

The US tech giant announced the two in-house chips at its Ignite conference in Seattle.

“At the scale we operate, it’s important for us to optimise and integrate every layer of the infrastructure stack to maximize performance, diversify our supply chain and give customers infrastructure choice,” said Scott Guthrie, executive vice president of Microsoft’s Cloud + AI Group.

It is another big-name client for Cambridge-based Arm, which designs and licenses advanced semiconductor technology that powers many of the world’s electronic devices.

Microsoft joins Apple in using Arm technology, with the iPhone maker using its processor designs to power custom M1 series computer chips.

In its first financial results post-IPO, Arm reported its highest-ever revenue of $806m (£656.5m). The results were in part thanks to the recent AI boom that helped it boost licence revenue by 106% year-on-year.

Supplying cloud chips to Microsoft should help bolster the number of chips it ships, which was 400 million less than predicted in the previous quarter’s results.

Mohamed Awad, SVP and GM of infrastructure business at Arm, said: “Together with leaders like Microsoft, we’re unlocking a new path to more efficiently building data centres with the performance required to deliver the definitive platforms and services of the coming decade.”