SK hynix is accelerating its memory manufacturing roadmap, targeting a threefold increase in wafer output by 2034 while advancing next-generation NAND technology to meet growing AI-driven demand.
SK hynix is ramping up its growth strategy with plans to significantly expand both DRAM and NAND production, underscoring its confidence in the long-term outlook for AI-driven memory demand.
According to reports, the company is targeting a threefold increase in wafer output by 2034, a much more aggressive goal than its previously announced plan to double capacity over the next five years. To support this expansion, SK hynix is accelerating development of its Yongin semiconductor cluster in South Korea, where four new fabs are planned. The first facility is expected to begin operations in 2027, with the overall timeline reportedly brought forward by nearly a decade.
The expansion is expected to boost the company’s DRAM production capacity from roughly 550,000 wafers per month today to around one million wafers per month by 2030. Chairman Chey Tae-won indicated that even these investments may struggle to keep pace with future demand, highlighting the rapid growth of AI and high-performance computing markets.
Alongside capacity expansion, SK hynix is advancing its technology roadmap. The company is reportedly preparing to begin mass production of its 375-layer NAND by the end of 2026, strengthening its position in the increasingly competitive high-density flash memory market. Future generations are already being planned, extending beyond 480 layers.
The developments come as the value of SK hynix’s investment in Japanese NAND maker Kioxia gains renewed attention amid a memory market recovery. While Kioxia remains a direct competitor, Chey said both companies continue to explore potential areas of cooperation outside core competitive segments.
Taken together, the accelerated fab investments, next-generation NAND roadmap, and rising value of strategic holdings position SK hynix to capitalize on the growing demand for advanced memory chips, particularly from AI infrastructure and data center applications.
