SK hynix plans to double its memory wafer capacity within five years, a move driven by what chairman Chey Tae-won called an AI-driven shortage that will persist until at least 2030. The announcement came at Computex in Taipei on June 2nd.
Chairman Chey Tae-won outlines massive production expansion at Computex Taipei event
The expansion targets the global supply of memory wafers used for DRAM and High Bandwidth Memory (HBM). SK hynix currently holds approximately 57% of the HBM market, a position that makes it a critical supplier to Nvidia's upcoming Vera Rubin platform. Customers are reportedly offering to buy EUV scanners and prefund fab lines due to near-zero available capacity.
Chey Tae-won declined to provide an exact dollar figure for the expansion, citing moving costs for land, equipment, and electricity as reasons why the buildout cost is difficult to pin down. Spending on the expansion in 2026 will exceed the $20 billion spent in 2025.
The lead time for a greenfield fab exceeds five years, placing fresh output near the tail end of the shortage window that Chey predicts. This timeline means new capacity will not alleviate the supply crunch before 2030.
AI data centers are currently swallowing the world's memory and storage supply, creating a perfect storm of demand and supply constraints. SK hynix is positioning its massive expansion to meet this sustained, high-end demand from the AI sector.
Discussion
0 comments
Log in to join the thread with a thoughtful take, question, or correction.