SK Hynix said it plans to double its wafer capacity over the next five years as it moves to relieve a global shortage of memory chips that remain crucial to artificial intelligence systems.

Chairman Chey Tae-won said the South Korean chipmaker is responding to a supply deficit in storage chips that he expects could continue until 2030. He made the comments in Taipei, where he said the company would increase spending to meet the imbalance between supply and demand.

The expansion is aimed at adding more output of memory wafers, the base material used in chip production. SK Hynix did not give a specific cost for the effort, with Chey noting that the company faces volatile prices for land, equipment and electricity. He said the firm would do what is necessary to finance the buildout.

The announcement underscores how persistent demand from AI data centers and related systems is straining memory supply chains. Memory chips are a core component in many AI products, and shortages have become a concern for the broader industry as companies race to expand computing capacity.

SK Hynix has already warned that the crunch may last for years, and Chey reiterated that view in his latest remarks. The company is among the largest suppliers of advanced memory chips and has benefited from strong demand tied to AI infrastructure spending. At the same time, the wider market has faced tight conditions across several inputs needed to expand manufacturing.

The planned increase in wafer capacity also signals that chipmakers may need to commit substantial resources well ahead of demand easing. Chey’s comments suggest SK Hynix is preparing for a prolonged period of constrained supply rather than a short-term cycle. The company’s investment push could help ease pressure on the market if it is able to bring additional capacity online as planned.

Memory chips have become increasingly important to the AI boom because they help store and move the data that large models require. That has made the segment a strategic focus for major semiconductor companies, which are competing not only on leading-edge logic chips but also on the components that support high-performance computing systems.

Chey did not outline the timing of individual projects or the locations for the new capacity. He also did not specify whether the expansion would come from existing sites, new plants or a mix of both. Still, his remarks point to a significant long-term commitment from one of the world’s most important memory suppliers.

For the chip industry, the plan adds another sign that manufacturers expect AI-related demand to remain elevated for years. If the shortage persists through the end of the decade as SK Hynix anticipates, more investment from major suppliers could become necessary to stabilize supply and support the next phase of growth in AI hardware.