Search

Microsoft and Intel to co-design faster Azure cloud servers for AI

Microsoft and Intel expanded their long-term collaboration to co-design parts of Azure infrastructure using EMIB packaging, CXL memory and next-generation Xeon CPUs.

Rows of server racks and equipment inside a data centre aisle.
Photo by Ismail Enes Ayhan on Unsplash

Microsoft and Intel will co-design faster Azure cloud servers for AI. In a press release dated 17 November 2026 they announced an expanded, long-term collaboration to advance select parts of Azure infrastructure using Intel Foundry's EMIB advanced packaging for Azure first-party silicon, multi-year Intel Xeon supply including Diamond Rapids CPUs, and joint work on Compute Express Link (CXL) memory, Intel QuickAssist Technology (QAT) and Intel Trust Domain Extensions (Intel TDX).

The companies say tests completed on 10 November 2026 showed that CXL memory combined with Intel Flat Memory Mode in Xeon 6 processors can deliver up to 2.3x greater performance and more flexible memory for demanding workloads in Azure M-series v4 virtual machines. A separate test completed on 2 November 2026 reported up to 2.5x faster SQL backups on Azure when using Intel QAT-enabled VMs. Those figures come from Microsoft and Intel's own tests and results may vary.

Briefly, EMIB is Intel's advanced multi-die packaging that lets different chip components sit closer together on the same package. Compute Express Link, or CXL, is an industry interconnect designed to let processors and accelerators share memory more efficiently. Intel Trust Domain Extensions, or TDX, helps protect virtual machine memory and CPU register state as confidential VMs move between hosts.

"Our customers need infrastructure built around the workloads that matter to them. Our longstanding collaboration with Intel brings together advanced packaging, custom compute, memory and security capabilities to support Azure compute and continued innovation across Azure. Together, we are connecting silicon and systems engineering to move the industry forward and meet the growing demand for Cloud + AI computing for our customers," said Saurabh Dighe, CVP, Azure System & Architecture, Microsoft.

I have followed this trend: when cloud engineers and chip designers work together they can tune processors, memory and software as a system. That can produce specialised VM instances tuned for AI, better memory efficiency for large models, and a tighter supply and launch sequence for new hardware.

For customers in Africa the practical effect depends on where Microsoft places the new hardware. Azure runs regions around the world, but not every new VM size or feature appears in every region at once. That can affect three concrete things for users in Africa: data residency (whether data and processing can stay in the same country or geography), latency (customers in regions without the new hardware may need to send workloads to other regions, increasing round-trip time), and costs (specialised hardware often appears first in a few regions and can carry different pricing or egress charges). The companies have not specified which Azure regions will get the EMIB-enabled packages or the earliest Diamond Rapids instances; customers should check region support before they deploy.

Read the full announcements from Microsoft and Intel. Before deploying, customers should check Microsoft's products-by-region page for current availability and any region-specific notes.

Newzroom dot Africa has reached out to Intel and Microsoft for comment and is awaiting a response. This story will be updated when they respond.

Source: Microsoft

Salute the work behind this story.

Blurbs

Up to 500 characters. Your name and location show with your blurb; your email is never shown. No links, email addresses or phone numbers in the blurb. Blurbs that break the terms of use are removed.

    More like this