SK hynix Plans Up to 13 Billion Dollars Investment in Advanced Packaging Lines in South Korea Alongside M15X Fab

Advanced packaging has rapidly become one of the most critical choke points in the AI semiconductor supply chain. As modern accelerators increasingly rely on tightly integrated memory and compute architectures, traditional manufacturing flows are no longer sufficient to meet demand. Technologies that allow high bandwidth memory to be stacked and closely coupled with compute dies are now fundamental to the product roadmaps of companies like NVIDIA and AMD, and that reality is forcing memory suppliers to rethink how vertically integrated they need to become.

In its latest announcement, SK hynix confirmed that it plans to invest heavily in advanced packaging capacity in Cheongju, South Korea, with total investment reaching up to 13 billion dollars. The official statement outlines the construction of a new advanced packaging fab known as P and T 7, designed to support global AI memory demand and improve end to end production efficiency.

According to the company, the decision reflects both market necessity and national industrial strategy. SK hynix described the investment as a strategic move aligned with South Korea’s policy of balanced regional growth, while also strengthening long term competitiveness and supply chain resilience. The P and T 7 facility will be built as an organic extension of the company’s upcoming M15X DRAM fab, effectively creating a tightly linked production ecosystem within the Cheongju Techno Valley Industrial Complex.

The practical implication is significant. Once DRAM and HBM products are manufactured at M15X, they can be transferred directly to the adjacent advanced packaging lines, where stacking, final processing, and integration will take place. This allows SK hynix to offer customers, particularly HBM buyers, a more streamlined one stop manufacturing solution rather than relying on fragmented external packaging flows.

This move also puts fresh focus on the broader question of packaging autonomy. Today, while SK hynix uses its own MR MUF vertical stacking process for HBM, full system level advanced packaging has largely depended on foundry partners, most notably TSMC and its CoWoS ecosystem. That dependency has become increasingly risky as demand for CoWoS capacity has outstripped supply, creating bottlenecks for AI accelerator vendors across the industry.

SK hynix has already signaled its intent to diversify packaging capacity geographically, with previous reports pointing to a planned 2.5D packaging investment in the United States valued at nearly 4 billion dollars. The far larger Korean investment suggests the company is now accelerating its efforts to strengthen packaging capabilities closer to its core memory manufacturing base.

What remains unclear is the exact technology path SK hynix will pursue at P and T 7. The company has not disclosed whether it intends to develop a fully proprietary advanced packaging solution, expand internal capabilities in partnership with established OSAT players like Amkor, or continue collaborating with foundry leaders by effectively bringing those processes in house through investment. Each option carries different implications for cost structure, control, and long term differentiation.

What is clear is the strategic direction. Advanced packaging is no longer an optional downstream step for memory suppliers. It is now a core competitive capability. By committing up to 13 billion dollars to P and T 7 alongside M15X, SK hynix is signaling that future leadership in the HBM and AI memory market will depend on delivering not just better dies, but complete, tightly integrated packaging solutions at scale.

Do you think SK hynix should prioritize building a fully proprietary advanced packaging technology, or is deepening partnerships with foundries and OSATs the safer path as AI demand continues to surge?