Microsoft and Intel said in releases dated Nov. 171 and Nov. 182 that Microsoft is the first company to announce use of Intel Foundry's EMIB packaging for Azure first-party silicon.

The companies say the agreement also confirms multi-year Xeon supply capacity, "including instances of next generation server CPUs, codenamed Diamond Rapids." The releases give no dollar value, contract length beyond "multi-year," Xeon volumes, or date for an Azure chip that uses EMIB. They name no Azure processor already built with it. This report relies on those releases alone, and they include no independent test results.

What the companies announced

Both releases describe an expansion of a "decades-long collaboration" under a "long-term agreement to advance select areas of Azure infrastructure." They spell out EMIB as Embedded Multi-die Interconnect Bridge and call it an advanced packaging technology from Intel Foundry.

The agreement also expands work across the Azure roadmap and Intel-based Azure hardware systems. Microsoft will collaborate with Intel on Xeon with Compute Express Link (CXL) memory, Intel QuickAssist (QAT) and confidential computing.

Intel's Naga Chandrasekaran, the company's EVP and general manager of Intel Foundry, offered the only on-record signal about scope. He said Intel would use "our advanced packaging capabilities with EMIB to scale future generations of Azure infrastructure," citing the "flexibility of heterogeneous integration." That wording points to future products rather than shipping ones.

Saurabh Dighe, CVP of Azure System & Architecture at Microsoft, said the collaboration "brings together advanced packaging, custom compute, memory and security capabilities." He gave no product or schedule.

Performance claims and their limits

The releases cite three items as existing memory and database work. Two carry performance figures, and the third, encryption of virtual machine memory and CPU register state as confidential VMs move between hosts using Intel Trust Domain Extensions, carries none.

The first is "up to 2.3x greater performance" in Azure M-series v4 virtual machines with CXL memory and Intel Flat Memory Mode in Xeon 6 processors. According to the footnote, the test compared a Standard_M76ds_2_v4 (76 vCPUs, 1,764 GB from 1 TB of DRAM plus 1 TB of CXL) with a DRAM-only Standard_M64lds_v4 (64 vCPUs, 1,007 GB). The workload was a sandbox microVM running a persistent Python interpreter with a 12 GB dataset. Testing was completed Nov. 10, 2026.

The second is "up to 2.5x faster SQL backups" on what the releases call the first QAT-enabled virtual machines in the cloud. The baseline is a two-socket Xeon Gold 8362 system, listed as ICX, with 1,024 GB of DDR4 running Windows Server 2025. It was compared with an Azure VM on Xeon 6973P-C (Granite Rapids) with 2.06 TB of memory running Windows Server 2025 Datacenter and SQL Server 2025. Testing was completed Nov. 2, 2026.

On the reading of these footnotes, neither comparison isolates the named feature. The 2.3x test differs in vCPU count and total memory as well as in CXL. The 2.5x test spans different processor generations and system configurations as well as QAT.

The releases do not say who ran the tests. Both footnotes state "results may vary."