The partnership between Intel and Microsoft spans more than three decades. It predates the cloud era entirely, and it has only intensified as enterprise workloads have grown more demanding. Intel served as the first reference architecture for SAP HANA, and that deep technical collaboration across Intel, Microsoft, and SAP continues to drive the capabilities customers rely on today.
As an in-memory database, SAP HANA operates extremely close to the hardware, consuming tremendous amounts of data in ways that demand tight optimization at the silicon level. Such proximity makes the collaborative relationship essential rather than optional.
At hyperscale, every efficiency gain compounds: every watt saved in a processor multiplies across millions of processes, and every improvement in how systems handle large data sets translates directly to cost and performance outcomes. Intel’s position in multi-socket scaling, the IP required to move from two sockets to four, eight, or 16, remains difficult to replicate, and it underpins the large-scale SAP deployments that Azure supports today.
At Microsoft Ignite 2025, Azure announced virtual machines (VMs) powered by Intel’s newest Xeon 6 processors, including support for CXL-based memory expansion.
Three of these developments land at a moment when SAP customers face a familiar set of pressures: in-memory databases consuming ever-larger footprints, mission-critical systems that cannot tolerate downtime, and finance teams scrutinizing every dollar spent on infrastructure.
1. CXL Technology Comes to the Cloud
A decade ago, running a VM with 32 terabytes of capacity and more than 1,700 virtual CPUs in a hyperscale cloud would have seemed far-fetched. Today, Azure supports VMs at that scale.
Getting there required sustained investment across all three partners, optimizing virtualization layers and ensuring reliability across an enormous number of components. The new Xeon 6 processors push the ceiling higher still. Intel’s 6500/6700 series, announced this year, supports scale-up configurations reaching up to 64 terabytes.
Equally significant is the introduction of CXL Flat Memory Mode, the first hardware-managed memory tiering system for CXL memory, now available in private preview on Azure. CXL technology decouples capacity scaling from compute scaling, allowing organizations to tailor configurations to actual workload requirements rather than overspending on compute simply to access the resources they need. For large S/4HANA deployments, this adds new flexibility in how memory resources can be configured. All this without compromising performance, thanks to the unique capabilities of Intel Xeon 6 and Intel Flat Memory Mode.
As the first hyperscaler to have announced such capability, Microsoft Azure continues its strong history of technology leadership with SAP infrastructure.
2. Broader VM Families Join the Lineup
SAP customers come in all sizes, and not every workload demands 32 terabytes of capacity or 1,700 vCPUs in a single VM. The Ignite announcements included new general-purpose and memory-optimized VM options based on Intel Xeon 6 processors—the Dlsv7, Dsv7, and Esv7 series—now in preview.
These deliver compute performance improvements of up to 15% compared to previous-generation Intel-based VMs, with the largest configurations scaling to 372 vCPUs and 2.8 terabytes of capacity. Networking bandwidth reaches 400 gigabits per second; remote storage throughput hits 800,000 IOPS.
The new VMs support three capacity-to-vCPU ratios, with options including and excluding local NVMe temp disks. General-purpose Dlsv7 and Dsv7 configurations suit a range of common workloads, from web-facing applications to backend application servers. Meanwhile, the capacity-optimized Esv7 family targets workloads with higher demands: large-scale databases (both SQL and NoSQL), analytics and business intelligence platforms, and in-memory systems like SAP S/4HANA.
3. Confidential Computing for Regulated Workloads
The collaboration between Intel and Microsoft extends to the security features built into the silicon itself, including transparent encryption and hardware-level isolation. Azure’s next-generation confidential VMs represent the latest development in that realm.
Now in preview with general availability expected in Q1 2026, these VMs are powered by fifth-generation Intel Xeon processors with Intel Trust Domain Extensions (TDX), enabling organizations to run confidential workloads without modifying application code. Data stays encrypted even during processing, addressing what had been a gap in end-to-end protection.
For customers in healthcare, defense, financial services, and other regulated industries, confidential VMs provide isolation enforced at the silicon level, keeping sensitive code and data encrypted while in use. Early adopters have deployed TDX-enabled VMs for secure collaboration platforms, end-to-end encryption architectures, and confidential AI environments that meet strict data sovereignty requirements.
The confidential VM offerings also include Intel Advanced Matrix Extensions (AMX), an accelerator built into the processor. AMX sets the foundation for emerging AI workloads that require hardware acceleration while maintaining confidentiality guarantees. Upcoming releases will add local NVMe SSD support, which allows for approximately five times the storage throughput and lower latency compared to the previous SCSI generation.
The Partnership Behind the Products
The announcements at Microsoft Ignite reflect years of joint development, but the partnership’s value to customers extends beyond any single product release.
Reliability is a case in point. The difference between managing a server with half a terabyte of capacity and one with more than 40 terabytes is more than only quantitative. At that scale, the sheer number of DIMM modules in play means error correction and fault compensation move from edge cases to central design concerns.
Intel has invested heavily in reliability, availability, and serviceability capabilities across Xeon processor generations, and the collaboration on testing and validation, tuning performance, stress-testing failure scenarios, and certifying configurations starts years before any processor reaches general availability.
Those fundamentals show up in customer outcomes. Migrations to Azure have produced performance increases of 30% to 40% across many systems, with analytics and business warehouse workloads improving by as much as 120% compared to older on-prem infrastructure.
Organizations that once maintained separate SAP systems by geography or business unit—often because infrastructure constraints made consolidation impractical—are now moving toward global S/4HANA instances. Reducing the number of systems under management generates dramatic cost savings.
Equally important is the assurance that growth paths exist. Committing to an architecture is easier when doubling business volume that does not require radical system changes. The sustained investment across Intel, Microsoft, and SAP means customers can plan for expansion knowing the infrastructure will keep pace with the needs of the business.