Intel Nears Multi-Billion-Dollar Deals for AI Chip Packaging Tech Amid Foundry Push

Intel's Foundry division is poised for a major breakthrough, with CFO David Zinsner announcing that the company is close to finalizing deals worth billions of dollars annually for its advanced packaging technology. Speaking at the Morgan Stanley Technology, Media and Telecom conference, Zinsner highlighted strong customer engagement, particularly for EMIB (Embedded Multi-die Interconnect Bridge), which offers design flexibility and cost efficiency compared to TSMC's dominant CoWoS method. This comes as AI chip demand overwhelms TSMC's packaging capacity, creating wait times of months to a year for clients like Nvidia, positioning Intel as a vital alternative.

The significance of these developments cannot be overstated for Intel, which has struggled with its Foundry business posting quarterly losses. Advanced packaging has emerged as a critical bottleneck in AI semiconductor production, where stacking multiple chips efficiently determines performance rather than just shrinking transistors. Intel's EMIB uses small silicon bridges for targeted interconnections, reducing expensive silicon use and supporting chiplet designs prevalent in modern GPUs. Facilities like Fab 9 in New Mexico and a new hub in Penang, Malaysia, are ramping up to meet this demand, bolstered by partnerships with AWS and Cisco. Geopolitically, Intel's U.S.-centric supply chain appeals to firms seeking to mitigate risks associated with overseas dependencies.

Adding momentum, CEO Lip-Bu Tan has shifted stance on the 18A process node, now open to external customers after initial plans limited it to internal use. Zinsner noted inbound interest in 18A-P, an upgraded version promising superior performance per watt versus TSMC's 2nm. While Foundry aims for break-even by late 2027, these wins could accelerate progress, though higher demand might require extra capacity investments.

Looking ahead, Intel anticipates announcing design wins ahead of schedule, potentially before mid-2026. Success here could challenge TSMC's packaging dominance and revitalize Intel's foundry ambitions, marking a pivotal shift in the AI chip ecosystem.