Microsoft is working on its own Arm processor designs for its datacenter servers, according to a Bloomberg report on December 18. Bloomberg also says Microsoft is exploring using another chip that would power some Surface PCs, the report adds.
While some are painting this as Microsoft responding to Apple’s recent decision to field its own Arm-based M1 processor, Microsoft and Qualcomm already had partnered since 2019 on Microsoft’s Arm chip that is inside the original Surface Pro X. The Pro X 2 uses the SQ2 chip, which is a variant of Qualcomm’s Snapdragon 8cx chip without 5G.
The part to me that’s more interesting is Microsoft using Arm in servers. Microsoft already had been working with Qualcomm and Caviium — along with Intel and AMD on Project Olympus, Microsoft’s next-generation cloud-hardware design it provided to the Open Comput Project. In 2017, Microsoft also announced that it has been involved with multiple ARM suppliers, including Qualcomm and Cavium on getting Windows Server to run ARM but for its own internal datacenter use only.
Back in 2017, I asked officials whether Microsoft would ever make Windows Server on ARM available externally to partners and customers. They said the technology was for internal use only for the purposes of evaluation of Azure services on Arm servers. But officials did note at the time that they believed Arm servers are good for internal cloud applications such as search and indexing, storage, databases, big data, and machine learning workloads.
I asked Microsoft officials about today’s Bloomberg report and received this response from Corporate Communications chief Frank Shaw:
“Because silicon is a foundational building block for technology, we’re continuing to invest in our own capabilities in areas like design, manufacturing and tools, while also fostering and strengthening partnerships with a wide range of chip providers.”
I found a fairly recent Microsoft job posting that mentions the work Microsoft has underway around ARM64 servers in its datacenters. My guess is this is the continuation of what the company announced back in 2017.
From that posting:
“Given the growth of our datacenters we are always investigating in new hardware and software solutions to meet our scale and customer demand. The Azure New Technology (ANT) team is looking ahead at future Cloud technologies, both hardware and software, and we are investigating and enabling those for our datacenter use. A good example of what we do is the development and deployment of ARM 64-bit Servers in our datacenters. This is a multi-year effort that consists of deep partnership engagements with multiple silicon companies, including collaborating on future hardware designs and software enablement. We are a very hands-on team with deep technical expertise in silicon, systems, operating systems (Windows and Linux) and application stacks.”
Will Microsoft continue to only use Arm servers for test purposes internally? Or will they start using Arm servers inside Azure datacenters to run new or existing Azure services sometime soon.
Another question (or three): Might Microsoft soon be following in AWS’ Gravitron footsteps sometime soon by making its own custom-built Arm cores available to its cloud customers? I asked Microsoft recently if the company had plans to provide Apple M1 instances in Azure to customers — similar to what AWS announced earlier this month — and was told by a spokesperson that “We are committed to meeting developers where they are and are always working to expand our offerings. We have nothing further to share today.”