Microsoft is reportedly developing two new chips for its cloud computing and artificial intelligence businesses. One chip is an Arm-based processor that will power the company’s Azure servers, while the other is a specialized chip that will improve the performance of its AI applications and services.
The move is part of Microsoft’s strategy to reduce its reliance on outside chipmakers such as Intel and Nvidia, and gain more control over the design and optimization of its hardware and software. Microsoft already uses its own chips for some of its products, such as Surface laptops and tablets, Xbox gaming consoles and HoloLens mixed reality headsets.
According to Bloomberg, which first reported the news, Microsoft’s Arm server chip will be similar to those that Amazon Web Services (AWS) and Apple have developed for their own cloud platforms and devices. Arm-based chips are more energy efficient and cost-effective than traditional x86 processors and can offer better performance for certain workloads. Microsoft has been working with Arm for several years to adapt its Windows operating system and software to run on Arm-based devices.
Microsoft’s artificial intelligence chip, on the other hand, will be a customized version of the Graphcore Intelligence Processing Unit (IPU), which is designed to accelerate machine learning tasks such as natural language processing, computer vision and recommender systems. Graphcore is a UK-based startup that has partnered with Microsoft since 2018 to provide its IPUs to Azure customers. Microsoft is also an investor in Graphcore and participated in its $200 million funding round in 2018.
By developing its own chips, Microsoft hopes to gain a competitive advantage over rivals in the cloud computing and artificial intelligence markets, which are expected to grow rapidly in the coming years. Microsoft is currently the second-largest cloud provider after AWS, with a 19% market share in the third quarter of 2020, according to Synergy Research Group. Microsoft is also a leader in AI research and innovation and has launched several initiatives such as Azure AI, Microsoft Research AI and AI for Good.
Microsoft has not officially confirmed or commented on its chip development plans, but the company is likely to reveal more details in the near future. The company has been hiring engineers and chip experts from the likes of Intel, AMD, Nvidia and Qualcomm, indicating that it is serious about its chip ambitions. Microsoft CEO Satya Nadella has also stated that the company’s vision is to create “the world’s computer,” which implies that it will need to have a strong presence in both the hardware and software domains.
Microsoft presents its own artificial intelligence and cloud chips at Ignite
Microsoft has been quietly working on its own custom chips to power its artificial intelligence and cloud computing services, and today it unveiled them to the world at its Ignite conference. The company claims these chips will offer superior performance and efficiency over existing solutions on the market.
The first chip is called Azure Maia AI accelerator and is designed to handle some of the most demanding generative AI tasks, such as Copilot, Microsoft’s AI-powered coding assistant. The Azure Maia chip can execute up to 100 trillion operations per second, making it one of the fastest AI chips in the world.
The second chip is an Arm-based CPU called Azure Cobalt and is optimized for running general-purpose cloud workloads such as web hosting, databases and analytics. According to Microsoft, the Azure Cobalt chip can deliver up to 50% more performance per watt than comparable x86 processors. Both chips are integrated into custom boards and servers that can be easily deployed in Microsoft data centers.
Microsoft says these chips are part of its vision to create a smarter, more sustainable cloud, and will enable new capabilities and experiences for its customers and developers. The company also says it will continue to invest in its own chip development, as well as partnering with other chipmakers, such as Intel, AMD and Nvidia, to provide the best solutions for its cloud services.
