Intel says Microsoft will use EMIB packaging for Azure chips
Microsoft plans to use Intel Foundry’s EMIB packaging for its own Azure silicon under an expanded infrastructure agreement that also secures multi-year Xeon supply. Chip identities, deployment dates and packaging economics remain undisclosed.
- Microsoft plans to use Intel Foundry’s EMIB packaging for its own Azure chips, Intel says.
- The agreement also confirms multi-year Xeon capacity, including upcoming Diamond Rapids processors.
- Neither chip-level deployment details nor evidence of EMIB-specific cost savings appears in the announcement.
Microsoft plans to use Intel’s advanced packaging for chips it designs for Azure, its cloud platform, under an expanded agreement announced November 18. The deal gives Intel Foundry another role in Microsoft’s infrastructure alongside a multi-year commitment to Intel server processors; it does not disclose the packaging work’s value or production schedule.

Chip testing at Microsoft’s Redmond lab, pictured in its November 2023 custom-silicon announcement. This historical photograph does not depict the newly announced EMIB designs. Photo courtesy of Microsoft. Source: Microsoft — custom silicon announcement.
A packaging commitment alongside processor supply
Intel said in its announcement that Microsoft is the first company to announce use of Intel Foundry’s Embedded Multi-die Interconnect Bridge, or EMIB, advanced packaging technology for Azure first-party silicon. First-party silicon means chips designed by Microsoft itself.
EMIB uses embedded bridges to connect separate chips within a package. Intel designs and manufactures semiconductors; Intel Foundry offers production services to other chip designers. The packaging commitment expands that customer relationship without establishing where the individual chips will be manufactured.
The long-term agreement also confirms multi-year supply capacity for Intel Xeon server CPUs, including next-generation processors codenamed Diamond Rapids. It extends work on Azure hardware, memory efficiency and security. Intel frames the combination as a way to improve performance, efficiency, scalability and resilience for cloud and AI workloads; the announcement does not quantify the agreement’s value or the capacity reserved.
For Microsoft, the relevant precedent is its move into designing more of the infrastructure it operates. In its November 2023 introduction, it presented Maia 100 as an accelerator for AI training and inference and Cobalt 100 as an Arm-based CPU for general-purpose cloud workloads. Microsoft described those chips as part of an effort to optimize infrastructure from silicon through servers, racks and cooling for internal and customer workloads. The company said it had begun custom-building servers and racks in 2016, with silicon becoming the missing component of that approach. Selecting packaging is another decision within that system design.

