SK Hynix’s Risky Pivot to DDR5 Tests Investors Ahead of $29 Billion Nasdaq Debut

SK Hynix reallocates HBM4 capacity to DDR5 to address shortages, risking AI market share as it plans a $29.4B Nasdaq listing and long-term pricing overhaul.

SK Hynix is trying to ride two horses at once. The South Korean chipmaker’s dominance in high-bandwidth memory (HBM) has powered a blistering 258.2% rally this year, but management is quietly shifting some capacity away from the next-generation HBM4 line to crank out more conventional DDR5 memory. The move addresses severe shortages in the traditional DRAM market, yet it risks ceding ground in the red-hot artificial-intelligence segment just as Nvidia prepares its next-generation Rubin platform.

The tension played out on the Seoul exchange last week. SK Hynix shares surged 10.88% on Friday to close at 2,425,000 won, trimming a weekly loss that still stands at 9%. Investors are digesting a strategic rethink that could reshape the company’s growth trajectory.

A $64 Billion Infrastructure Bet

The company isn’t scaling back its overall ambitions. SK Hynix has committed roughly 100 trillion won — equivalent to about $64 billion — to new fabrication facilities, with the bulk earmarked for its Yongin semiconductor cluster. A fourth plant inside that complex is slated to begin operations in 2033. To help fund the expansion, the chipmaker is heading to the United States: American depositary receipts will start trading on the Nasdaq on July 10, 2026, with the order book opening on July 6 and a final price set on July 9. The listing is expected to raise up to $29.4 billion.

Alongside the capital raise, SK Hynix is tearing up the old playbook on pricing. Long-term supply contracts will no longer carry price ceilings, and the average contract length has been extended to five years. The logic is straightforward: in a market where supply is tight and demand is surging, removing caps lets the company harvest maximum margin. The first-quarter 2026 operating margin of 72% shows the strategy is already paying off, with net profit soaring and revenue jumping 198% year on year to 52 trillion won.

Dominance and Its Discontents

SK Hynix controlled 58% of the global HBM market in the first quarter of 2026, and its traditional DRAM share sits at a stable 29%. Analysts at UBS project the company will capture 70% of the HBM4 orders for Nvidia’s upcoming Rubin platform, a forecast buttressed by a formal partnership signed in June. The entire 2026 HBM output has already been sold out, and industry experts expect the memory segment to grow 30% this year.

Yet the cyclical nature of the chip business casts a long shadow. The industry is notorious for boom-bust cycles, and the current wave of capacity expansion — not just by SK Hynix, but also by rivals Samsung and Micron — raises the specter of a supply glut after 2026. If the AI infrastructure buildout decelerates sooner than expected, the new factories could become an enormous drag on profitability.

The DDR5 Dilemma

The decision to divert resources from HBM4 toward DDR5 adds another layer of risk. DDR5 shortages are real, and capturing that demand protects a large revenue base. But every wafer assigned to conventional memory is one less for the high-margin HBM line, where technical leadership is essential. HBM production is notoriously challenging — stacking multiple memory layers and connecting them at microscopic scale creates high defect risks. A slip in yields or a slower ramp of HBM4 could let Samsung or Micron claw back market share, eroding the pricing power SK Hynix has worked so hard to establish.

The new no-cap pricing strategy is a double-edged sword. It magnifies profits while demand outpaces supply, but it also amplifies losses if the market turns. Such aggressive terms demand an unbroken run of strong orders from a handful of hyperscaler customers. Any meaningful budget cut from those key clients would hit SK Hynix directly, and geopolitical strains in the region add another unpredictable variable.

What Comes Next

For the moment, the bull case still prevails. The structural shortage of AI memory chips supports the investment thesis, and the Nasdaq listing will provide a fresh valuation benchmark. Investors will watch the IPO pricing closely on July 9 for a read on global appetite. Then comes the second-quarter earnings report due shortly after the listing. If record margins hold, last week’s dip will quickly become a footnote.

But the window for complacency is narrow. If HBM4 volume production suffers delays, if the AI spending spree cools, or if competitors close the technology gap, the same leverage that lifts SK Hynix today could just as easily pull it down. The pivot to DDR5 is a calculated hedge, but it also marks a moment when the company’s famous focus begins to blur.