SK Hynix Chief Executive Kwak Noh-jung has warned that the global memory chip industry is on course for its most severe supply shortage in history in 2027, saying demand from artificial intelligence customers is expected to outstrip production capacity for years despite massive industry investment.
The comments, made in an interview with Reuters on Friday, provide one of the strongest indications yet that leading semiconductor manufacturers expect the AI infrastructure boom to remain intact well beyond the current investment cycle. They also challenge recent investor concerns that spending on AI data centers could be nearing a peak.
Speaking on the day SK Hynix began trading on Nasdaq, Kwak said supply constraints, rather than weakening demand, are likely to become the defining challenge for the memory industry over the remainder of the decade.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
“We forecast that next year will be the worst year in the industry’s history from the supply perspective,” Kwak said.
“Our customer demand continues to go up, while our capacity has limitations,” he added. “We still forecast that customer demand will remain higher than our supply capacity even beyond 2030. But we are doing our best to solve the problem.”
SK Hynix has emerged as one of the biggest beneficiaries of the artificial intelligence revolution, becoming the world’s leading supplier of high-bandwidth memory (HBM), a critical component used alongside AI accelerators produced by Nvidia and other chipmakers.
HBM enables AI processors to rapidly access and process the enormous volumes of data required to train and run large language models and other advanced AI applications. As hyperscale cloud providers and technology companies race to expand AI computing capacity, demand for HBM has surged far faster than manufacturers can build new production facilities.
Unlike conventional memory chips, HBM is significantly more complex to manufacture, requiring advanced packaging technologies, sophisticated production processes and close collaboration with GPU manufacturers. Expanding capacity therefore takes years rather than months, creating persistent supply bottlenecks.
Kwak’s comments suggest that even with aggressive investment across the industry, production will struggle to keep pace with customer demand through the end of the decade.
The warning came as SK Hynix made a strong debut on the Nasdaq, reflecting investors’ confidence in the company’s strategic position within the global AI supply chain.
Shares rose 13.3% to $168.85 during Friday afternoon trading, underscoring continued investor interest despite recent volatility across semiconductor stocks.
The listing provides SK Hynix with greater access to U.S. capital markets and broadens its visibility among international institutional investors at a time when AI-related companies remain among the most closely watched stocks globally.
U.S. Manufacturing Remains An Option
Kwak also said the company continues to evaluate the United States as a potential location for future wafer fabrication facilities, although no final investment decision has been made.
The comments come as semiconductor manufacturers diversify manufacturing footprints in response to geopolitical tensions, government incentives and efforts by customers to strengthen supply chain resilience.
Building wafer fabrication plants in the United States would align with broader efforts by Washington to expand domestic semiconductor manufacturing through incentives provided under industrial policy initiatives.
SK Hynix has already committed substantial investments in the U.S. The company is spending approximately $4 billion to build an advanced chip packaging facility in Indiana, where high-bandwidth memory will be integrated with advanced AI processors.
It has also announced plans to invest $10 billion to establish an AI solutions business in the United States, broadening its activities beyond semiconductor manufacturing into AI-related technologies and services. Those investments are intended to deepen relationships with major U.S. technology companies, many of which are among SK Hynix’s largest customers.
Investors Question Sustainability of AI Spending
Kwak’s optimistic outlook comes against a backdrop of growing investor debate over whether the AI infrastructure boom is approaching a turning point.
In recent weeks, semiconductor stocks have experienced heightened volatility as investors reassessed expectations for capital spending by large technology companies.
Concerns have been fueled by reports that Apple is exploring greater use of Chinese semiconductor suppliers for parts of its supply chain and that Meta is considering commercializing excess AI computing capacity to generate additional returns on its infrastructure investments.
Those developments have prompted questions about whether hyperscale companies may eventually moderate the pace of AI-related spending after investing hundreds of billions of dollars in new data centers.
However, industry executives continue to believe that current demand remains substantially higher than available supply.
But the expected shortage has important implications for the semiconductor market. When demand consistently exceeds supply, manufacturers typically enjoy stronger pricing power, improving profit margins and supporting long-term earnings growth.
Memory prices have already risen sharply this year as cloud computing providers and AI developers compete for limited supplies of advanced HBM products. Industry analysts expect that continued shortages could sustain elevated pricing for several years, benefiting leading manufacturers such as SK Hynix, Samsung Electronics and Micron Technology.
At the same time, prolonged supply constraints could slow deployment of AI infrastructure by making advanced memory more difficult and expensive to obtain.
Capacity Expansion Faces Practical Limits
Although semiconductor companies are investing tens of billions of dollars in new manufacturing facilities, expanding production is constrained by several factors. Constructing advanced fabrication plants requires significant capital, specialized equipment, highly skilled engineers, and long lead times that often exceed three years.
In addition, advanced memory production depends on complex supply chains involving semiconductor manufacturing equipment, specialty chemicals, substrates, and advanced packaging technologies, all of which have experienced periodic bottlenecks.
These constraints explain why even aggressive investment may not eliminate shortages in the near future.
However, Kwak’s comments have bolstered the view held by many industry leaders that artificial intelligence remains in the early stages of a multi-year investment cycle rather than approaching its conclusion. His forecast that demand will continue exceeding supply beyond 2030 suggests SK Hynix expects AI adoption to broaden well beyond today’s hyperscale cloud providers into enterprise computing, autonomous systems, robotics, healthcare, industrial automation, and other sectors.
The remarks thus provide a sort of reassurance that recent weakness in semiconductor shares may reflect short-term concerns about valuations and market positioning rather than a deterioration in the industry’s underlying fundamentals. If SK Hynix’s projections prove accurate, the company and the broader memory industry could remain central beneficiaries of sustained global investment in AI infrastructure for the remainder of the decade.