SK Hynix accelerates AI-driven memory chip expansion with $25 billion investment in new factories

South Korean semiconductor giant SK Hynix unveils a major $38.8 billion investment to build new DRAM and NAND factories, aiming to meet soaring demand driven by artificial intelligence and data infrastructure needs, with plans to start production by 2028 and 2029.

SK Hynix has announced a large expansion of its memory-chip business as it prepares for a longer period of demand from artificial intelligence systems and data infrastructure. The South Korean company said it will build two new factories, one for DRAM and one for NAND flash, in a project worth 35.2 trillion won, or about $25 billion, for the DRAM plant and 19.1 trillion won, or about $13.5 billion, for the NAND facility. The company expects the NAND line to start operating in December 2028, followed by the DRAM plant in June 2029.

The investment is intended to address a market that remains tight even before the new capacity comes on stream. SK Hynix said, citing Omdia research, that demand for DRAM and NAND is expected to rise at a compound annual growth rate of 19% through 2030. In that context, the company said its ability to supply large volumes at the right time will be a key competitive advantage.

Tom’s Hardware reported that the new spending is part of a broader three-year push worth about $74.6 billion, adding to a separate $90 billion mega fab complex already under construction in Gyeonggi Province. TechTarget said about 80% of the planned AI-related investment is expected to go into high-bandwidth memory, or HBM, the stacked memory used in AI accelerators to feed data quickly to processors. That aligns with SK Hynix’s role as a major supplier to AI chip makers, including Nvidia.

The company is also trying to position itself beyond conventional memory production. According to a separate report from Tom’s Hardware, SK Hynix and SanDisk recently unveiled the High Bandwidth Flash specification, an open standard that combines the capacity of 3D NAND with the performance goals of HBM for AI inference workloads. SK Hynix has also announced a nearly $4 billion packaging and research facility in Indiana, underlining how the group is spreading its AI memory strategy across manufacturing, advanced packaging and R&D.

Disclaimer: This content is intended for informational purposes only. Readers are advised to exercise their own judgement, conduct due diligence, or consult a qualified expert before acting on any information provided.