SK Hynix CEO Kwak Noh-jung warned that artificial intelligence demand will create the worst supply shortage in memory chip industry history by 2027, according to a Reuters interview. The chipmaker cited surging demand from AI data center buildouts as the primary driver, with customer orders outpacing production capacity.
SK Hynix has become central to the AI chip supply chain through its leadership in high-bandwidth memory, a specialized component used in GPU clusters and AI accelerators. The shortage is expected to persist beyond 2030, with demand continuing to exceed what manufacturers can deliver.
Memory chip shortages have occurred before in the industry's history, most notably during the 2022-2023 cycle when oversupply followed pandemic-era buildouts. The current projection differs in scope: Hynix's statement frames 2027 as a supply constraint driven by new demand rather than a cyclical swing. The company is among the world's largest memory chip makers, competing directly with Samsung and Micron for contracts with hyperscalers building AI infrastructure.
Data center operators have accelerated orders for high-bandwidth memory as they expand GPU clusters. Supply bottlenecks in specialty memory could affect the timeline and cost of AI infrastructure deployment globally. Hynix did not specify production capacity targets or forecast a timeline for when supply would balance demand.