Arm Ltd., whose chip designs can be found in everything from smartphones to data centers, will move most of the electronic design automation applications used by its engineers to Amazon Web Services Inc.’s cloud platform.

The project was detailed by AWS this morning. 

Modern processors contain billions of transistors organized into logic gates, which are in turn organized into more complex components such as caches and artificial intelligence accelerators. It’s impossible for chip engineers to work out every single detail by hand. As a result, they automate parts of the workflow using specialized applications known as electronic design automation or EDA tools, which are the central focus of Arm’s newly announced cloud deal with AWS.

The chip designer will move the bulk of its EDA workloads to the Amazon.com Inc. subsidiary’s platform as part of the project. Already, it relies on AWS to run simulations that help its engineers test new semiconductor blueprints. The firm also uses the cloud platform to host a deployment of Databricks Inc.’s machine learning toolkit, which it uses to analyze data collected on product development efforts to find areas for improvement.

Some of Arm’s own chip technology is helping to support its cloud migration efforts. The firm’s AWS environment includes EC2 compute instances powered by Graviton2, a central processing unit the Amazon.com subsidiary has developed based on Arm’s Neoverse processor design.  

“Now that we can run on Amazon EC2 using AWS Graviton2 instances with Arm Neoverse-based processors, we’re optimizing engineering workflows, reducing costs and accelerating project timelines to deliver powerful results to our customers more quickly and cost effectively than ever before,” said Rene Haas, president of Arm’s IP Products Group.

According to AWS, the chip designer has managed to provide a sixfold performance boost for some EDP workflows while improving the throughput of certain other engineering workloads by a factor of 10. The returns from switching to the cloud have proven so compelling that Arm now plans to reduce its data center infrastructure by at least 45%, which will include a 80% cut in the amount of computation performed on-premises.

The semiconductor industry in which ARM competes presents sizable revenue opportunities for cloud providers such as AWS. Developing a new chip involves numerous steps, from algorithmically checking circuit reliability to optimizing the manufacturing process, that in many cases require considerable computing power

Another factor likely helping cloud infrastructure demand in the industry is that the last few years have seen the emergence of numerous new venture-backed chip startups. The public cloud’s pay-as-you-go-model can be more practical for them than the buying and maintaining on-premises hardware. 

Image: AWS