Following a stock price surge of over 400% in the past year and record-breaking profits, SK Group Chairman Chey Tae-won has committed to expanding production capacity for AI memory chips to meet soaring demand driven by a global data center construction boom. Speaking at a conference in Washington on February 20, the leader of South Korea's second-largest conglomerate referred to High Bandwidth Memory (HBM) as a "monster chip" that is generating massive profits for SK Hynix. The chipmaker's stock performance has led the global semiconductor industry, fueled by explosive growth in demand for AI chips.

In January, SK Hynix announced a significant increase in capital expenditure for 2026 compared to the previous year, aiming to meet demand for HBM chips required by AI accelerators designed by companies like NVIDIA. The company's entire production capacity for memory chips in 2026 has already been fully booked.

This capacity expansion plan underscores how the global AI race is reshaping semiconductor supply dynamics. U.S. tech giants from Microsoft to Meta are projected to invest approximately $6.5 trillion this year in AI infrastructure, a record level of spending that is causing a worldwide shortage of memory chips.

Market leadership is being further solidified. The global memory chip market is currently dominated by SK Hynix, its South Korean rival Samsung Electronics, and U.S.-based Micron Technology. In the critical AI component segment of HBM, SK Hynix has established a leading position. With its 2026 capacity already sold out, Micron is experiencing similar demand for its HBM products.

American technology firms are engaged in a large-scale infrastructure competition to gain an edge in AI technology development. From Microsoft and Meta to other industry giants, total investments in AI-related infrastructure are expected to reach around $6.5 trillion this year. This unprecedented investment wave is directly driving demand for high-performance memory chips.

The HBM chips produced by SK Hynix serve as core components in AI accelerators from companies like NVIDIA and AMD, used for training and operating artificial intelligence services. As AI models continue to grow in scale, demand for these high-performance memory chips is experiencing explosive growth.

While profit expectations are soaring, risks remain. Analysts continue to raise earnings projections for SK Hynix. According to Chairman Chey, the average analyst forecast for the company's 2026 operating profit has increased from approximately $50 billion at the end of last year to $70 billion in January, with some institutions even raising their expectations to over $100 billion.

However, Chey simultaneously issued a warning. "This sounds like good news," he stated, "but it could also turn into $100 billion in losses." The chairman cautioned that the potential for future losses remains due to rapid technological iterations that could dramatically alter the competitive landscape.

This cautious statement reflects the cyclical nature of the semiconductor industry and uncertainties brought by technological transformation. Despite current strong demand, industry history shows that supply-demand imbalances and technological transitions can rapidly change market dynamics.

Energy infrastructure is emerging as a new bottleneck. Chey also highlighted growing infrastructure challenges, revealing that SK Group is exploring building power plants adjacent to AI data centers since failing to meet energy demands could have "catastrophic" consequences.

This statement underscores new constraints facing the AI chip industry. As AI models and data centers continue to expand, power supply is becoming another critical bottleneck following chip manufacturing capacity. For chipmakers planning significant capacity expansions, ensuring stable energy supply has become a strategic priority.

Although Chey did not specify the scale of SK Hynix's capacity expansion, the company has clearly stated it will substantially increase capital expenditure. This commitment demonstrates that despite existing technological and market risks, SK Hynix maintains confidence in the long-term demand outlook for AI chips.