TW StockArticle is online

Microsoft and Intel Expand Partnership to Build a More Differentiated Azure Infrastructure

數字匠人
Microsoft and Intel Expand Partnership to Build a More Differentiated Azure Infrastructure

Table of Contents




You might want to know



  • How will Microsoft and Intel’s expanded cooperation help Azure meet rising demand for cloud computing and AI?

  • What improvements could CXL memory, Intel QuickAssist, advanced packaging, and confidential computing bring to Azure customers?



Main Topic


Intel (英特爾, INTC-US) and Microsoft (微軟, MSFT-US) have announced a long-term agreement that expands their decades-long relationship. The companies plan to work together on important areas of Azure infrastructure as demand for cloud computing and artificial intelligence continues to grow. As part of the announcement, Microsoft disclosed for the first time that it will use Intel Foundry’s EMIB™ (Embedded Multi-die Interconnect Bridge) advanced packaging technology in a custom Azure chip.


The agreement brings together Intel’s chip and packaging capabilities with Microsoft’s cloud engineering expertise. According to Intel, the collaboration establishes a multi-year capacity supply for Intel® Xeon® processors, including the next-generation server CPU codenamed Diamond Rapids. It also broadens joint work on Azure’s technology roadmap and on Azure hardware systems built around Intel architecture.


For cloud infrastructure, the agreement addresses more than processor supply. As systems become more complex and workloads increasingly depend on AI, data processing, and large memory capacity, performance can be shaped by the way processors, memory, packaging, and software work together. The companies’ stated direction is to coordinate these parts of the system rather than treat each component as an isolated upgrade.


One area of focus is Intel Xeon processors that support Compute Express Link (CXL). Microsoft and Intel will collaborate on these processors to improve Azure memory efficiency and advance system-level innovation. CXL is relevant to infrastructure designs that need to connect computing resources and memory effectively. In Azure, this work is intended to help address demanding workloads in which memory capacity and access can influence overall performance.


Microsoft also plans to work with Intel on Intel QuickAssist (Intel QAT) and confidential computing. These initiatives build on Azure’s existing memory and database performance work. The companies describe them as complementary efforts: acceleration technologies can help improve the processing of particular tasks, while confidential computing can help protect sensitive information as it is handled within cloud infrastructure.


Azure M-series v4 virtual machines equipped with CXL memory, together with Intel Xeon 6 processors using Intel Flat Memory Mode, are reported to deliver up to 2.3 times the performance and memory flexibility for performance-intensive workloads. This is a maximum improvement stated for the specified configuration and workloads; it should not be interpreted as a universal result for every application. Actual outcomes depend on factors such as workload characteristics, software, and system configuration.


Another performance claim concerns SQL backups. Microsoft says Azure virtual machines are among the first in the cloud industry to support Intel QuickAssist technology. With this capability, SQL backup speeds on Azure can increase by up to 2.5 times, supporting improvements in performance and reliability. For organizations that regularly back up large or business-critical databases, faster backup operations may help reduce the time and resources needed for that process, although results can vary by environment.


Security is also part of the collaboration. Through Intel Trust Domain Extensions (Intel TDX), confidential virtual machines can maintain encryption for virtual machine memory and CPU register state when they are migrated between hosts. Protecting these states during migration is intended to help preserve the confidentiality of sensitive workloads as infrastructure operations move virtual machines from one host to another.


The central idea is system-level co-optimization: Microsoft and Intel are aligning chips, packaging, compute, memory, and software to address Azure’s cloud and AI requirements as a connected infrastructure challenge. This approach also links near-term improvements, such as memory efficiency and database backup speed, with longer-term infrastructure development, including custom chips and advanced packaging.


Intel’s EMIB technology is a notable element of the announcement because Microsoft says it will use the packaging approach in an Azure custom chip for the first time. Advanced packaging can support the integration of multiple components within a package. In the companies’ view, combining that flexibility with large-scale manufacturing capabilities can help develop future Azure infrastructure for AI-era workloads. The announcement does not specify the custom chip’s full design, release timing, or detailed performance characteristics.


The investment is intended to give Microsoft customers an infrastructure platform that is more predictable, scalable, and resilient. The companies aim to optimize computing capacity for cloud and AI workloads that require high performance, improve memory efficiency, and strengthen protection for sensitive workloads. These are stated goals of the partnership; the announcement does not provide specific service availability dates or pricing details.


Saurabh Dighe, Microsoft corporate vice president of Azure Systems and Architecture, said customers need infrastructure tailored to their critical workloads. He described the long-standing Microsoft–Intel relationship as combining advanced packaging, customized compute, memory, and security technologies to support Azure compute and ongoing platform innovation. Dighe also emphasized bringing chip and system engineering together to respond to growing customer demand for cloud and AI computing.


Naga Chandrasekaran, Intel executive vice president and chief technology and operations officer, and general manager of Intel Foundry, said the companies share a clear view of AI infrastructure development. As systems grow more complex, he said, they need to be designed through coordinated optimization across chips, packaging, compute, memory, and software. Chandrasekaran added that using Intel EMIB advanced packaging to extend future generations of Azure infrastructure could combine the flexibility of heterogeneous integration with the large-scale manufacturing capacity required in the AI era, with the aim of delivering better performance and cost effectiveness for Azure customers.


Together, the commitments span processor availability, memory architecture, acceleration, security, and packaging. Their significance will ultimately depend on how these technologies are implemented in Azure services and how well they perform across real customer workloads. The announcement nevertheless signals a broadening of the partnership from processor supply toward a more integrated approach to cloud infrastructure design.



Key Insights Table



































Aspect Description
Long-term cooperation Microsoft and Intel are expanding their decades-long relationship through a long-term agreement focused on Azure infrastructure and growing cloud and AI needs.
Processor capacity The agreement includes multi-year Intel Xeon processor capacity, including the next-generation server CPU codenamed Diamond Rapids.
Advanced packaging Microsoft says it will use Intel Foundry’s EMIB™ technology in an Azure custom chip for the first time.
Memory and performance Azure M-series v4 virtual machines with CXL memory and Intel Xeon 6 processors using Intel Flat Memory Mode can provide up to 2.3 times the performance and memory flexibility for demanding workloads.
Database acceleration Microsoft says Azure virtual machines supporting Intel QAT can increase SQL backup speeds by up to 2.5 times, with potential benefits for performance and reliability.
Confidential computing Intel TDX helps keep confidential virtual machine memory and CPU register state encrypted during migration between hosts.


Afterwards...


The next stage will be to understand how these infrastructure commitments translate into Azure services, customer access, and performance across different workloads. Useful areas for further exploration include how CXL-based memory configurations behave under varied application demands, how processor and packaging choices affect efficiency, and how acceleration technologies influence database operations in practical deployments.


Confidential computing also remains an important area for continued development. Protecting memory and processor state during migration is one part of a wider security challenge that includes workload isolation, key management, software assurance, and operational controls. Clear technical documentation and testing across different use cases can help customers assess how these protections fit their own security requirements.


More broadly, the partnership reflects a shift toward designing cloud platforms as integrated systems. Future advances will depend on coordinating hardware, packaging, memory, software, and manufacturing while also measuring performance, energy use, reliability, and cost. Continued work on transparent benchmarks and workload-specific evaluations will help determine where these technologies deliver the greatest practical value. As cloud and AI demands evolve, system-level collaboration may become increasingly important to building infrastructure that is adaptable, secure, and capable of scaling with customer needs.

Last edited at:2026/11/20

數字匠人

Author

數字匠人

Idle Passerby

A easy-going and enjoys walking in the scenery of life. He is also an observer and sees the world with a calm heart.