SK hynix Reportedly Evaluates Intel EMIB as CoWoS Capacity Tightens the AI Packaging Market
SK hynix is reportedly exploring Intel’s advanced packaging technology as pressure continues to build across the AI supply chain, with fresh reporting suggesting the memory giant is working with Intel on early 2.5D packaging research centered on Embedded Multi die Interconnect Bridge, or EMIB. According to ZDNet Korea, cited by multiple market reports, SK hynix is evaluating whether Intel’s EMIB substrate approach could be used to connect high bandwidth memory with logic chips as demand for AI packaging outpaces the industry’s current supply. At this stage, the reported work appears to still be in an early research and validation phase rather than in confirmed volume deployment.
The timing makes strategic sense. Advanced packaging has remained one of the most important bottlenecks in the AI hardware boom, especially as TSMC’s CoWoS capacity continues to run under heavy demand from leading accelerator and AI platform customers. Reuters recently reported that SK hynix is already investing heavily in advanced packaging capacity in South Korea to support rising HBM demand tied to artificial intelligence, underscoring just how central packaging has become to the company’s long term growth plans.
What makes the Intel angle especially interesting is that EMIB is not being positioned only as a backup option, but as a technology that could become more relevant as hyperscalers and chip designers seek alternatives to crowded CoWoS capacity. Market reports this month have pointed to growing investor interest in Intel’s packaging business, with reporting that Intel is in discussions with external partners for high value advanced packaging work and that EMIB is gaining visibility as a credible alternative in certain AI packaging scenarios.
According to the Korean reporting summarized in other market coverage, SK hynix is specifically studying whether EMIB could be used to integrate HBM with logic dies and is also reviewing the materials and components that would be needed if the technology eventually moved closer to mass production. That is an important distinction because it suggests the company is not simply benchmarking Intel’s technology on paper. It is reportedly examining what a real manufacturing path could look like if EMIB proves viable for future HBM packaging use. Still, neither Intel nor SK hynix has publicly confirmed a production deal here, so the most accurate reading for now is that this is a reported technical evaluation, not a finalized commercial packaging shift.
The broader industry backdrop also supports why this kind of exploration is happening now. AI customers are scaling faster than parts of the backend semiconductor ecosystem can comfortably handle, and packaging has become one of the most critical choke points in the stack. If SK hynix is indeed validating Intel EMIB as another packaging path, it would signal that even top tier AI memory suppliers are actively looking beyond the standard TSMC centered route to protect future supply flexibility. That would be a meaningful development not just for Intel’s foundry and packaging ambitions, but for the entire AI hardware market, where backend manufacturing capacity is increasingly as important as leading edge silicon itself.
For Intel, the report fits neatly into Lip Bu Tan’s current message around expanding the company’s role beyond CPUs and into advanced packaging and foundry services. While EMIB is not a direct one for one substitute for every CoWoS deployment, the fact that major industry names are even being linked to it in active evaluation shows Intel may be gaining a more serious foothold in one of the most strategically valuable parts of the AI supply chain. Whether that turns into volume business will depend on yields, cost structure, technical fit, and customer confidence, but the direction of travel is becoming harder to ignore.
Do you think Intel’s packaging business can become a true second pillar in the AI supply chain, or will TSMC remain too far ahead for competitors to meaningfully break its grip?
