Intel Secures Major Microsoft Order for Maia 2 AI Chip Production
Intel has landed a significant contract with Microsoft, marking a pivotal moment for Intel Foundry as it gains the confidence of major external clients. According to industry analyst Charlie Demerjian of SemiAccurate, Microsoft has chosen Intel Foundry to manufacture its upcoming Maia 2 chip, utilizing the advanced 18A or 18A-P process node.
Advancements in Intel’s 18A-P Process Technology
The 18A-P node represents a substantial upgrade over Intel’s standard 18A process. It incorporates the second generation of RibbonFET and PowerVia technologies, which are designed to deliver superior performance and energy efficiency. Key improvements include newly engineered low-threshold voltage components, optimized elements to minimize leakage, and refined ribbon width specifications. These enhancements are particularly valuable for AI accelerators deployed in data centers, where maximizing performance-per-watt is critical due to the scale of deployment.
Microsoft’s Transition from TSMC to Intel Foundry
Microsoft’s first-generation Maia 100 chips were previously manufactured using TSMC’s N5 process, featuring CoWoS-S interposer technology. These chips are notable for their 820 mm² die size, 500 W TDP (with a maximum design power of 700 W), and integration of 64 GB HBM2E memory delivering 1.8 TB/s bandwidth. The Maia 100 also includes 500 MB of L1/L2 cache and achieves peak tensor performance of 3 PetaOPS at 6-bit precision, 1.5 PetaOPS at 9-bit, and 0.8 PetaFLOPS for BF16. Connectivity is robust, with 600 GB/s backend network bandwidth via twelve 400GbE ports and 32 GB/s host bandwidth through PCIe Gen 5 x8.
With the Maia 2 project, Microsoft is considering relocating future production of its Maia accelerators to Intel Foundry. If the Maia 2 initiative demonstrates high yield and rapid turnaround, Microsoft may continue leveraging Intel’s more advanced process nodes for subsequent generations.
Future Prospects: Advanced Intel Nodes and Packaging
Looking ahead, Microsoft could adopt Intel’s forthcoming 18A-PT and 14A nodes for future AI and high-performance computing (HPC) applications. The 18A-PT process is specifically tailored for AI and HPC workloads that demand sophisticated multi-die architectures. This node introduces advanced packaging features, including a redesigned back-end metallization layer, through-silicon via (TSV) capabilities for vertical interconnects, and support for state-of-the-art hybrid bonding interfaces with competitive pitch dimensions. These innovations enable highly scalable chiplet integration, positioning Microsoft to take advantage of advanced packaging technologies in future AI accelerator designs.
Intel’s collaboration with Microsoft on the Maia 2 chip underscores the growing importance of advanced semiconductor manufacturing processes and packaging solutions in meeting the performance and efficiency demands of next-generation AI hardware.
