SK Hynix Aims to Double Wafer Capacity Within Five Years
2026-06-03
On June 2, 2026, at the Taipei International Computer Exhibition (Computex 2026), Chey Tae-won, Chairman of South Korea's SK Group, announced a bold plan: SK Hynix will double its total wafer capacity over the next five years.
Decision Driven by Supply-Demand Imbalance
The direct catalyst for this capacity-doubling decision is the long-term supply-demand imbalance triggered by the explosive growth of AI infrastructure. As the core supplier of HBM for NVIDIA's GPUs, its HBM production capacity for 2026 is already completely sold out, and tight supply is expected to persist into 2027. Chey Tae-won explicitly stated that the global shortage of memory chips is projected to last until 2030. As tech giants like Meta, Google, and Microsoft compete to invest heavily in building AI data centers, and with the proliferation of NVIDIA's new generation of AI PCs, demand for High Bandwidth Memory (HBM), server DRAM, and NAND flash is growing exponentially. He admitted, however, that the company has not yet precisely calculated the total investment required. SK Hynix will "mobilize all necessary resources" to advance expansion regardless of cost, despite facing rising costs for equipment, land, utilities, and other challenges.
Roadmap for Doubling Capacity
SK Hynix's five-year blueprint to double capacity is built on a clear production strategy, progressing through a dual-track approach: "building new super-bases" and "upgrading existing fabs."
Cheongju M15X Fab: World's First Dedicated HBM4 Base. As the vanguard of the expansion, the M15X fab in North Chungcheong Province, with a total investment exceeding 20 trillion won, commenced operations ahead of schedule in Q4 2025 and began mass production of critical HBM4 products in February 2026. Its monthly capacity is planned to ramp up from an initial 10,000 wafers to 55,000-60,000 wafers by the end of 2026. It is projected to account for over half of the global incremental capacity for HBM4.
Yongin Semiconductor Cluster: Core Capacity Hub for the Next Decade. The Yongin complex in Gyeonggi Province is the strategic centerpiece of this expansion, with a total planned area of 4.9 million square meters for four large-scale fabs. The Phase 1 fab (Y1) has already secured 31 trillion won in investment, and its operational start date has been moved up from May 2027 to February 2027. Upon full completion, the Yongin cluster is designed to become the world's largest production base for AI memory, specifically built to meet the explosive demand from AI servers and data centers over the next decade.
Existing Fab Upgrades & Overseas Expansion. SK Hynix is also converting part of its traditional DRAM capacity to HBM processes by upgrading existing fabs like M10 and M14 in Icheon. Additionally, its advanced packaging facility in Indiana, USA, is expected to be completed in 2028, bolstering the global HBM supply chain.
Challenges Ahead
If successfully executed, this five-year plan would dramatically increase SK Hynix's monthly DRAM wafer capacity from the current approximately 500,000 units, potentially surpassing Samsung Electronics (about 700,000 units) to become the global leader. However, the expansion path is not without obstacles. Chey Tae-won acknowledged that building a new fab takes at least three years, and greenfield projects from scratch require over five years, making this a decadelong endeavor. Furthermore, SK Hynix faces significant challenges, including long lead times for critical equipment like ASML's EUV lithography machines, constraints on domestic power and water resources in South Korea, and the relentless pressure to maintain its technological leadership.
Semicone IC Independent Distributor
Semicone is a global independent distributor of electronic components. Specializing in MOSFETs, driver ICs, switching interface components, ADCs, power chips and other categories of IC components, it is committed to providing original, high-quality products, efficient logistics and supply solutions to global ODM/OEM/EMS customers. The following are today's selected chips.
● FDP083N15A-F102: An N-channel PowerTrench® MOSFET from ON Semiconductor, with a 150V voltage rating and 83A high current capacity.
● DRV8718SQRVJRQ1: An automotive-grade 8-channel intelligent half-bridge gate driver from Texas Instruments, integrating a current detection amplifier.
● TMP708AIDBVR: A fully integrated resistive programmable temperature switch from Texas Instruments, with a high typical accuracy of ±0.5°C and a low static current of only 40µA.
● AD7705BRUZ-REEL: A 16-bit Σ-Δ analog-to-digital converter (ADC) from ADI, with a built-in programmable gain amplifier, supporting dual-channel differential input.
● IRF540NSTRLPBF: A HEXFET® power MOSFET from Infineon, achieving extremely low on-resistance (44mΩ @10V) using advanced technology.
Email: info@semicone.com
Apple Farewell to Vision Pro, All-in on AI Smart Glasses
Nikon's Lithography Strategy Adjustment: Breaking Through with Cost Advantages
Related Article
NSING and ADI launch second 2026 price. Amid AI supply chain strain, NSING raises 10%–20% and ADI targets 30% on military-grade chips.
From MCUs to Analog Chips, NSING and ADI Raise Prices Again
Sony and TSMC seal a $4.7B joint venture in Kumamoto to build next-gen CMOS image sensors. Sony takes 60% control, TSMC 40%, targeting 2029 mass production.
Sony and TSMC Form Joint Venture to Focus on CMOS Image Sensors
