HBM dominant SK Hynix predicts that the shortage of storage supply will continue until 2030, and plans to double its production capacity within five years.
SK Hynix will double its wafer manufacturing capacity to alleviate the shortage of storage chips.
In the field of HBM, SK Hynix, the global storage chip giant with the highest market share, plans to double its storage chip wafer production capacity in the next five years to meet the continuously rising demand for storage chips brought about by the frenzy of artificial intelligence data center construction. The chairman of the SK Group, Cui Taiyuan, told reporters at the 2026 Computex conference in Taipei, China that the storage giant is preparing for the possibility of severe and continuous shortage of storage chips that could last until 2030. Although SK Hynix intends to significantly increase investment to expand HBM/DRAM/NAND storage chip capacity, Cui Taiyuan did not disclose how much money the company plans to spend.
SK Hynix, along with Samsung Electronics and Micron Technology, Inc. based in the United States, jointly dominate the global storage chip market supply and demand landscape, and have always been the major beneficiaries of the booming data center construction driven by the exponential expansion of artificial intelligence training/inference computing power. These two South Korean storage chip giants both consider NVIDIA Corporation as one of their largest customers.
Meta Platforms, Alphabet Inc. Class C, and Microsoft Corporation, among other North American tech giants, have recently collectively announced larger-scale AI capital expenditure plans to continue expanding their AI data centers. This has led to a continued surge in demand for advanced storage chip products, helping SK Hynix and Micron's market value to surpass the epic $1 trillion mark for the first time last week.
Following Samsung Electronics into the trillion-dollar market value club, another South Korean storage giant, SK Hynix, also saw its total market value surpass $1 trillion last week, joining Samsung Electronics in driving the Korean stock market benchmark index, the KOSPI composite index, to repeatedly hit new highs. The core driving force behind this is the near-endless demand for storage and the expectation of rising storage chip prices driven by AI training/inference. With the soaring stock prices of the world's two largest storage chip manufacturers, Samsung Electronics and SK Hynix, the Korean benchmark index, the KOSPI, has nearly doubled in less than 6 months this year, making the near 80% increase in the entire year of 2025 seem insignificant in comparison.
SK Hynix to double capacity in five years, supply shortage to continue until 2030
SK Hynix is embarking on a historic expansion of its storage chip production capacity, planning to double its memory chip wafer production capacity over the next five years to meet the strong global demand for storage chips driven by the construction of global artificial intelligence infrastructure. Chairman Cui Taiyuan pointed out at the Taipei Computex conference that the global storage chip supply gap is expected to persist until 2030, indicating that SK Hynix needs to increase capital expenditure on expansion and capacity to address the long-term supply-demand imbalance by purchasing more ASML Holding NV ADREUV lithography machines, cutting-edge HAR etching and film deposition equipment, etc. This further cements its absolute core position in the global storage market and demonstrates its forward-looking layout for HBM memory and NAND storage demand in the AI era.
Whether it's Alphabet Inc. Class C's massive TPU AI computing cluster or NVIDIA Corporation's large-scale AI
GPU computing clusters, they all rely on HBM storage systems that need to be fully integrated with AI chips. With current technology giants accelerating the construction of new or expanded AI data centers, there is a growing demand for server-grade DDR5 storage and enterprise-grade high-performance SSDs/HDDs. Samsung Electronics, SK Hynix, and Micron Technology, Inc. are all involved in the most core areas of storage: HBM, server-grade high-performance DRAM (including DDR5/LPDDR5X), and high-end data center-level SSDs. They are the most direct beneficiaries in the "AI memory + storage stack," benefiting from the "super dividends" of AI infrastructure.
In the High Bandwidth Memory (HBM) market, SK Hynix maintains a leading position with a market share of 58%, far exceeding competitors Samsung and Micron - Samsung and Micron each hold about 21%. The company clearly positions itself as the core HBM supplier for NVIDIA Corporation's Vera Rubin AI computing cluster system, and plans to expand its cooperation with more partners to cover a wider range of data center and AI computing projects.
By deeply integrating AI supercomputing and cloud computing service provider ecosystems, SK Hynix's product and market strategies have closely aligned with industry-leading players, reinforcing its strategic position as a key supplier of global AI infrastructure.
From cyclical commodities to strategic core assets, the LTA agreement reshapes the stock price and profit logic of storage chip manufacturers
Under the surge in AI computing demand, the fundamentals of the storage chip industry are undergoing unprecedented structural reshaping. The traditional cyclic storage market was dominated by consumer electronics and inventory clearance cycles, with prices and demand fluctuating greatly.
In the current super cycle of storage chips, the persistent and near-endless demand for high-bandwidth memory (HBM), server DRAM, and enterprise-grade SSDs as driven by AI training and inference has elevated storage to a strategic core asset of AI infrastructure. The global total addressable market (TAM) for storage chips is expected to reach approximately $1.7 trillion by 2028, indicating a fundamental shift from being a commodity to an indispensable AI computing infrastructure asset.
Morgan Stanley expects that DRAM revenue could increase from $143 billion in 2025 to $636 billion in 2026, reaching approximately $1.237 trillion by 2028, while NAND revenue could soar from $71 billion to $454.5 billion.
Goldman Sachs Group, Inc.'s core assessment is that the valuation framework of the storage industry is shifting from the traditional price-to-book ratio (P/B) to price-to-earnings ratio (P/E). In the past, storage was seen as a cyclical commodity with concerns about profits collapsing rapidly after prices peak, leading to long-term undervaluation. However, this cycle is being jointly driven by AI servers, HBM, enterprise-grade SSDs, long-term supply agreements (LTA), and constrained capacity, significantly improving profit visibility. Goldman Sachs Group, Inc. reportedly raised target prices for Samsung, SK Hynix, and Kioxia, increasing the target price for Samsung to 480,000 Korean won, SK Hynix to a maximum of 3.5 million Korean won, and Kioxia to 93,000 Japanese yen. The logic behind this is that leading storage companies should not be priced as "bottom-of-the-cycle assets," but should be valued closer to stable cash flow-generating AI infrastructure assets.
Morgan Stanley's bullish logic for storage chip manufacturers is even more aggressive: this super cycle of storage will be "higher and longer," as AI demand has spread from GPUs to CPUs, ASICs, inference servers, and intelligent autonomous computing systems. GPUs handle parallel matrix computing, but intelligent autonomous AI requires CPUs to handle task scheduling, state management, tool calls, and API execution, significantly increasing demand for server DRAM, HBM, and enterprise-grade SSDs. According to their framework, the value of storage in the capital expenditure of cloud service providers has jumped from just over ten percentage points in the early stages of AI to possibly over 50% this year, with a further expected increase to nearly 73% by 2030; this implies that storage is no longer just a complementary component of servers, but a core resource that determines the throughput, latency, energy efficiency, and scalability of AI systems.
Related Articles

In April, the number of job vacancies in the United States increased significantly to the highest level in nearly two years, supporting the hawkish expectations of the Federal Reserve on the resilience of the labor market.

Stocks have been rising for two straight months! Wall Street is not buying into this "reluctant bull market" in the US.

The embodied intelligence track remains hot! SoftBank plans to increase its stake in Siasun Robot & Automation.
In April, the number of job vacancies in the United States increased significantly to the highest level in nearly two years, supporting the hawkish expectations of the Federal Reserve on the resilience of the labor market.

Stocks have been rising for two straight months! Wall Street is not buying into this "reluctant bull market" in the US.

The embodied intelligence track remains hot! SoftBank plans to increase its stake in Siasun Robot & Automation.






