SK Group Chairman Chey Tae-won announced at Computex 2026 in Taipei on June 2 that SK Hynix plans to double its memory chip wafer production capacity over the next five years — the company’s first public disclosure of a specific production target. Chey told reporters the AI-driven memory shortage will persist until at least 2030, reaffirming a forecast he first made in March, and said the company would commit whatever capital is required to hit that goal. Capital expenditure in 2026 will climb well above the 30.2 trillion won (about $20 billion) spent in 2025, he said, though he declined to name a precise figure given volatility in land, equipment, and electricity costs. “Whatever we need, we’ll provide it,” Chey said. “Until 2030 there’s still some shortage.” He also confirmed SK Hynix has filed to list American depositary receipts in New York this year, which would rank among the largest U.S. debuts by a foreign company.
The announcement comes as SK Hynix has emerged as the clearest beneficiary of the AI infrastructure boom. The company holds roughly 57–58% of the global HBM market and about 32% of all DRAM supply; its 2026 HBM capacity is fully sold out, and 2027 capacity is already locked up by Nvidia, Google, and Microsoft under long-term agreements. SK Hynix’s market cap crossed $1 trillion last week — up over 260% year-to-date — as hyperscaler data center spending drove both it and Micron past the milestone. Q1 2026 operating profit reached 37.6 trillion won, up 405% year-over-year, on a 72% operating margin. However, as Tom’s Hardware notes, a greenfield fab takes more than five years to bring online, meaning fresh capacity would arrive near the very end of the shortage window Chey himself is projecting — doing little to relieve the near-term crunch. Chey added that he wants SK Hynix to become the primary HBM supplier for Nvidia’s next-generation Vera Rubin platform, and said the company needs to expand manufacturing partnerships beyond TSMC.