Redmond, WA - November 15, 2023 - Microsoft today announced two major advancements in its hardware portfolio: the Azure Maia AI accelerator chip and the Azure Cobalt Arm-based CPU. These new products are designed to deliver exceptional performance and efficiency for a wide range of workloads, from AI and machine learning to general-purpose cloud computing.
The AI chip, called the Azure Machine Learning Compute (AMLC) chip, is designed to accelerate AI workloads such as training and inference. It is based on a new architecture that Microsoft says will deliver significant performance improvements over existing AI chips. The cloud CPU, called the Azure Data Compute (ADC) processor, is designed to handle big data workloads such as data analytics and machine learning. It is also based on a new architecture that Microsoft says will deliver significant performance improvements over existing cloud CPUs.
Microsoft says that both the AMLC chip and the ADC processor will be available in Azure cloud services in early 2024. The company is also making the chips available to partners who build their own AI and cloud infrastructure.
“At the scale we operate, it’s important for us to optimize and integrate every layer of the infrastructure stack to maximize performance, diversify our supply chain, and give customers infrastructure choice,” Microsoft’s executive VP, Scott Guthrie, said in a statement.
The launch of the AMLC chip and the ADC processor is a major milestone for Microsoft in its efforts to compete with Amazon Web Services (AWS) and Google Cloud Platform (GCP) in the cloud computing market. AWS and GCP both offer their own custom-designed AI and cloud CPUs, and Microsoft is now joining the fray with its own chips.
![]()
The availability of the AMLC chip and the ADC processor will give Microsoft's Azure cloud services a significant performance advantage over AWS and GCP for AI and big data workloads. This could help Microsoft attract more customers to Azure and gain market share from AWS and GCP.
The AI chip is also designed and tested with the group at OpenAI.
“We were excited when Microsoft first shared its designs for the Maia chip, and we’ve worked together to refine and test it with our models,” Sam Altman, Open AI CEO, said in a statement. “Azure’s end-to-end AI architecture, now optimized down to the silicon with Maia, paves the way for training more capable models and making those models cheaper for our customers.”