SEOUL – SK Hynix outlined plans to quicken spending on advanced memory chip capacity after reporting record earnings, reflecting surging demand for artificial intelligence (AI) and a drive to stay ahead of rivals.
The South Korean company notched a bigger-than-expected 68 per cent jump in operating income to 9.21 trillion won (S$8.6 billion) in the June quarter on a 35 per cent surge in revenue. It now aims to accelerate planned investments pre-emptively in areas including high-bandwidth memory (HBM), the company said.
Demand for memory is expected to continue to grow, with customers launching new products in the second half of 2025. The emergence of ChatGPT is prompting nations around the world to build their own AI infrastructure and data that are independent, the company said.
The results reinforce SK Hynix’s lead over bigger rival Samsung Electronics in HBM that help Nvidia’s AI processors.
SK Hynix has benefited from Samsung’s delay in obtaining Nvidia’s certification for its most advanced product, 12-layer HBM3E. That’s given SK Hynix an unusually long lead time in the highly lucrative segment.
Now, SK Hynix is racing to stay ahead of deeper-pocketed Samsung. Some investors expect the chips-to-smartphone company – which in July reported its first quarterly profit decline since 2023 – to have hit bottom over the summer. Nvidia is shifting towards a new generation of memory chips, offering Samsung an opening.
SK Hynix’s shares have shed about 10 per cent from a 2025 high they notched in July. Goldman Sachs downgraded the chipmaker for the first time in more than three years on expectations of competitors encroaching on its market share. SK Hynix and Samsung together account for more than 80 per cent of the HBM market.
Client Nvidia, which controls more than 90 per cent of the market for chips needed to build AI systems. But long-term prospects for advanced memory makers may depend on expanding beyond Nvidia, given the incentives on the US company to diversify its vendors and negotiate lower prices.
SK Hynix is looking to strengthen ties with the likes of OpenAI and deepen ties by building AI infrastructure. OpenAI chief executive Sam Altman said in a post on X on July 21 that his company will bring online one million graphics processing units (GPUs) by the end of the year and is trying to figure out how to scale that a hundredfold.
SK Group chairman Chey Tae-won, accompanied by SK Hynix CEO Kwak Noh-jung recently met Mr Altman at OpenAI’s office in San Francisco. They discussed collaboration on semiconductors including HBM, according to local media. BLOOMBERG