South Korean memory semiconductor manufacturer SK Hynix recently completed the largest U.S. share sale by a foreign company in history, raising $26.5 billion as its American depositary receipts began trading on Nasdaq.
The high-profile listing, which saw shares surge 13% on the first day of trading and valued the enterprise at $1.2 trillion, provides critical capital to fund an unprecedented domestic expansion strategy.
Driving this complex financial maneuver is the AI sector’s massive appetite for advanced memory, which is prompting global chipmakers to commit hundreds of billions of dollars to new manufacturing capacity.
SK Hynix intends to deploy its newly raised capital directly into manufacturing facilities, advanced manufacturing equipment, and heavy machinery. This capital injection directly supports a 1,100 trillion South Korean won (~$733 billion) mid-to-long-term corporate investment roadmap designed to solidify the company’s leading position in the memory market.
View All Kwak Noh-Jung, CEO of SK Hynix, emphasized the urgency of these infrastructure projects. “The current supply shortage, coupled with surging demand, has made expanding production capacity essential.” He further noted the structural shift in the technology landscape, stating, “The AI industry has moved beyond the training phase and entered an era in which AI services are being deployed at scale.”
Capital-intensive global competition
The aggressive capital raises and infrastructure expansions by SK Hynix place intense pressure on its primary competitors, which are simultaneously executing massive capital projects of their own. U.S.-based Micron Technology recently announced an acceleration of its domestic manufacturing investments, committing more than $250 billion through 2035 to support facilities in New York and Idaho.
Micron’s CEO Sanjay Mehrotra framed the expansion as historically significant. “As America celebrates its 250th anniversary, data and memory are foundational to the modern economy—and Micron is increasing our U.S. investments to more than $250 billion through 2035 to meet that moment.” Internationally, Micron also broke ground on a $9.3 billion expansion at its Hiroshima facility in Japan, targeting next-generation DRAM and high-bandwidth memory (HBM) output by 2028.
Meanwhile, Samsung Electronics is advancing its own manufacturing timeline, preparing to fully ramp its P4 fabrication plant in Pyeongtaek by late 2026, with subsequent phased expansions for its P5 facilities planned through 2030. The industry consensus is that these rapid expansions are necessary to capture market share in a fast-evolving technological landscape.
Analyzing AI demand
Market observers note that this sector-wide expansion is fundamentally tethered to the explosive growth of AI applications. In an email to EE Times, Dwight Morse, a principal solutions architect at SiliconExpert, commented, “The main driver behind memory demand has been artificial intelligence. It started with HBM supporting the training of models and has spread to NAND for storage of tokens. It’s likely that these facilities will be operating at full capacity for the foreseeable future.”
This view is echoed by broader industry tracking, which indicates that legacy hardware configurations are rapidly being replaced. In emailed responses to EE Times, Clark Tseng, senior director of industry research and statistics at SEMI, said capital expenditures are a direct response to structural market shifts. “AI-related memory demand is not only driving new investment but also consuming a disproportionate share of both existing and new memory capacity, particularly for HBM and server storage,” Tseng said.
Market vulnerabilities and supply chain
Despite the optimistic capital expenditure, the memory semiconductor industry faces substantial macroeconomic and operational hurdles. The Bank for International Settlements (BIS) recently issued a stark warning regarding the potential for an AI infrastructure spending bubble.
The institution cautioned that if the financial returns on AI tools fail to match the massive capital outlays, a sudden pullback by technology firms could severely disrupt the global financial system. In the near term, while AI servers demand high-end memory, consumer markets are showing signs of fatigue. TrendForce reports that record-high contract prices have pushed consumers in the PC and smartphone markets to their affordability limits, leading to more moderate price increases across standard components.
There are also growing concerns about oversupply. Because semiconductor fabs require at least three years from groundbreaking to reach yield stabilization, the current wave of multi-billion-dollar investments by SK Hynix, Samsung, and Micron will largely translate into output around 2028.
However, Tom Hsu, an analyst at TrendForce, offered EE Times a more measured perspective on this timeline. “Supported by robust AI-driven demand, TrendForce expects the market to absorb most of the projected supply growth in 2028,” Hsu stated by email, adding that renegotiations of long-term agreements would only occur in the event of a severe downturn, which is not their base-case scenario. SEMI’s Tseng also noted that manufacturers still have room to adjust their equipment installation and ramp pace depending on how demand conditions evolve.
Geopolitical tensions have introduced additional complications to this expansion race. Both SK Hynix and Samsung are reportedly restructuring their supply chains to eliminate dependencies on Chinese equipment manufacturers, such as Mattson Technology. This strategic pivot aims to mitigate the risk of disruption if Chinese suppliers are added to the U.S. Department of Commerce’s Entity List, forcing the South Korean companies to rapidly secure alternative providers for critical photoresist and thermal processing equipment.
Corporate leadership maintains that the structural shift toward AI fundamentally alters historical boom-and-bust memory cycles. SK Group chairman Chey Tae-won dismissed concerns of an impending downturn following the Nasdaq debut.
“We’re going to double our capacity within five years, and all my customers say, ‘That’s not enough, man. We need more,’” Chey stated. Addressing the cyclical nature of the business, Chey added that the proliferation of automated technologies ensures sustained hardware consumption. “The AI agent, a physical AI robot, actually needs a lot of memory chips,” he noted, expressing confidence that the demand for memory has permanently changed.
See also:
[SK Hynix Plans $713B Domestic Investment](https://www.eetimes.com/sk-hynix-plans-713b-domestic-investment/)
[ASML Raises Outlook, Plans More EUV Capacity](https://www.eetimes.com/asml-raises-outlook-plans-more-euv-capacity/)
[TSMC Boosts 2026 Expansion Budget, Adds $100B to U.S. Investment](https://www.eetimes.com/tsmc-boosts-2026-expansion-budget-adds-100b-to-u-s-investment/)
[Kioxia All Set to Raise the NAND Game in AI SSDs](https://www.eetimes.com/kioxia-all-set-to-raise-the-nand-game-in-ai-ssds/)