SK Hynix Commits $38 Billion to New Memory Chip Plants as AI Demand Outpaces Supply
South Korean chipmaker SK Hynix has approved a $38.1 billion investment to build two new memory chip factories, one of the largest single capacity expansions in the industry’s history, as demand for the components powering the global AI boom continues to outstrip supply.
The company’s board signed off on the investment Friday, committing 35.2 trillion won ($24.9 billion) to a new DRAM facility known as Yongin Y2, and a further 19.1 trillion won ($13.2 billion) to a NAND flash plant in Cheongju, designated M17.
A shortage reshaping the entire chip industry
The investment lands amid a genuine supply crunch. Memory prices have surged sharply this year as AI data centers absorb enormous volumes of high-bandwidth memory (HBM) and other advanced chips, components critical to training and running large AI models. The shortage has pushed SK Hynix and rival memory makers Samsung and Micron to record share-price gains, as investors bet the supply imbalance will persist for years rather than months.
Counterpoint Research vice president Neil Shah said the fresh investment won’t affect near-term output, since the new plants are built for 2029 and beyond, but reflects how seriously chipmakers are taking the durability of AI-driven demand.
Racing against a rival, and a self-imposed deadline
SK Hynix currently ranks as the world’s second-largest memory maker, but faces intensifying competition from Samsung, which reclaimed the top spot by market share in the DRAM segment during the second quarter of this year, according to Counterpoint Research.
The Yongin Y2 facility is the second of four planned fabs at SK Hynix’s broader Yongin Semiconductor Cluster, a manufacturing campus roughly 20 miles south of Seoul. Construction on Y2 begins in July 2027, with its first cleanroom expected online by June 2029. Notably, SK Hynix has compressed its original timeline for completing all four Yongin fabs from 2045 down to 2033, a twelve-year acceleration reflecting how urgently the company sees the opportunity in front of it.
The Cheongju NAND facility, by contrast, builds on an existing manufacturing base, the site already houses three SK Hynix fabs along with established power and water infrastructure, allowing construction to move faster there than at an entirely new site.
Part of a much larger national push
Friday’s announcement is one piece of a considerably larger commitment. SK Hynix has outlined plans to invest up to 600 trillion won across the Yongin cluster alone, alongside a further 100 trillion won earmarked for its other manufacturing base in Cheongju, all part of a broader South Korean government-backed push to grow the country’s domestic chip sector, an effort that also includes rival Samsung.
The ripple effects reach ordinary consumers
The memory shortage isn’t confined to data centers. Apple has faced supply pressure on some MacBook Air, Mac mini, and Mac Studio configurations tied directly to the same memory squeeze, with Apple CEO Tim Cook describing the shortage and resulting price increases as comparable to a “100-year flood.” The strain illustrates how the same underlying AI-chip demand driving investments like SK Hynix’s is now feeding through into consumer electronics pricing and availability well beyond the data center.
The dynamics echo what Infineon reported in its own record Q3 earnings this month, where AI-related power semiconductor demand, tied to the same broader data center boom, has become the company’s fastest-growing business line, underscoring how thoroughly AI infrastructure spending is reshaping the global chip industry from multiple angles at once.
