By Wen-Yee Lee, Max A. Cherney and Heekyong Yang
TAIPEI, June 2 (Reuters) – SK Hynix aims to double its wafer capacity over the next five years, the chairman of parent SK Group said on Tuesday, as AI-driven demand places the South Korean memory-chip maker at the heart of an industry boom.
Chey Tae-won was speaking at the Computex conference in Taipei, where executives from some of the world’s top technology companies, including Nvidia, are gathering.
“We are going to double the whole capacity over the next five years … there are a lot of obstacles and hurdles, but we will get over them and expand,” Chey told reporters.
The SK Group chief also reiterated his view that memory supply bottlenecks could persist through 2030, maintaining a forecast he first made in March.
SK Hynix, Nvidia’s leading supplier of high-bandwidth memory (HBM) chips that are crucial for AI applications, held a 58% share in the global HBM market in the first quarter, followed by Samsung Electronics and Micron Technology, each with a 21% share, according to Counterpoint Research.
Chey said that Nvidia’s upcoming AI personal computer architecture would require substantial amounts of memory, supporting long-term demand growth.
He also said he hopes his company can be a major HBM supplier for Nvidia’s technologically advanced Vera Rubin system.
Some analysts say the AI boom is reshaping the traditionally cyclical memory industry.
Goldman Sachs raised its 2028 operating profit forecasts for SK Hynix and Samsung by 24% and 23.3%, respectively, to 454 trillion won ($299.62 billion) and 610 trillion won, citing sustained AI-driven demand.
Last week, SK Hynix topped $1 trillion in market value for the first time, joining Samsung and Micron in reaching the milestone on an AI-driven rally.
HBM COMPETITION INTENSIFIES
Competition in the HBM market has intensified as memory makers race to secure a larger share of the fast-growing AI infrastructure market.
Samsung unveiled on Tuesday a mock-up of its future HBM5 chip for the first time and introduced its Heat Path Block (HPB) thermal management technology, which will be incorporated into the new product.
Last week, Samsung said it had begun shipping samples of its latest HBM4E chip to customers, putting it ahead of rivals in distributing an advanced memory product critical for AI data centers.
About the roadmap for SK Hynix’s HBM4E chips, Chey said it would depend on customer demand. “There’s only one customer for HBM4E right now,” he said, referring to Nvidia.
The company also needs more partnerships in Taiwan, not just with TSMC, the world’s largest contract chipmaker, Chey added.
Asked whether prices for DRAM and HBM chips would continue to rise, Chey said the industry should focus on sustainable growth, adding that excessive price increases could hurt the broader AI ecosystem and undermine long-term industry development.
“The whole AI industry needs more sustainability,” he said. “We have to continue to grow, but sudden jumps in prices can become a problem and actually hurt sustainability.”
(Reporting By Wen-Yee Lee and Max A. Cherney in Taipei and Heekyong Yang in Seoul; Editing by Anne Marie Roantree and Muralikumar Anantharaman)
Comments