--- title: "Experience the shock brought by the \"storage supercycle\"! Cloud giants are not only competing for HBM but also rushing to help SK Hynix purchase lithography machines for expansion" type: "News" locale: "en" url: "https://longbridge.com/en/news/285733928.md" description: "Large technology companies such as Microsoft, Google, and Amazon are actively competing for SK Hynix's HBM memory chips, proposing unprecedented investment plans, including purchasing advanced chip manufacturing equipment to expand production capacity. This move reflects the severity of the global memory chip shortage, especially against the backdrop of surging demand for artificial intelligence. Goldman Sachs predicts that memory chip prices will rise significantly by 2026, exceeding previous optimistic expectations" datetime: "2026-05-08T13:23:39.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/285733928.md) - [en](https://longbridge.com/en/news/285733928.md) - [zh-HK](https://longbridge.com/zh-HK/news/285733928.md) --- # Experience the shock brought by the "storage supercycle"! Cloud giants are not only competing for HBM but also rushing to help SK Hynix purchase lithography machines for expansion According to the Zhitong Finance APP, media reports citing informed sources indicate that HBM storage giant SK Hynix is being aggressively pursued by major global tech companies in an "alternative" manner—Microsoft, Google, and Amazon, the world's largest cloud computing giants, have unprecedentedly proposed large-scale investments in its new production lines and plan to personally fund the purchase of increasingly expensive chip manufacturing equipment, including ASML lithography machines, advanced HAR etching, and thin-film deposition equipment, to expand capacity. This is part of a collaborative plan to secure as much capacity and supply as possible amid their competition to ensure HBM, DRAM/NAND storage chip supply. This type of bidding and investment in capacity is unprecedented in the global memory chip industry, highlighting the severe shortage of such components worldwide. Amid the unparalleled AI boom driving a surge in demand for computing infrastructure, memory chip manufacturers are generally struggling to keep pace with the exponential expansion in demand. Memory chips are critical components in the large-scale new AI data centers being built globally, as well as in smartphones, personal computers, wearable electronics, and other electronic products. Whether it is Google's massive TPU AI computing cluster or NVIDIA's enormous AI GPU computing cluster, both rely on HBM storage systems that fully integrate AI chips. Additionally, current tech giants accelerating the construction or expansion of AI data centers must purchase server-level DDR5 storage and enterprise-level high-performance SSD/HDD in large quantities. Samsung Electronics, SK Hynix, and Micron Technology are simultaneously positioned in these three core storage areas: HBM, high-performance server DRAM (including DDR5/LPDDR5X), and high-end data center-level SSDs, making them the direct beneficiaries of the "AI memory + storage stack," effectively reaping the "super dividend" from the AI infrastructure wave. Regarding the price increase of DRAM/NAND storage chips, Wall Street financial giant Goldman Sachs recently assessed that—**the price increase for storage in 2026 will far exceed the optimistic expectations previously provided by the institution. Goldman Sachs has recently raised its forecast for DRAM storage chip price increases from about 150% to 250%-280%, and the forecast for NAND price increases from about 100% to 200%-250%. In other words, Goldman Sachs believes this is not an ordinary inventory correction cycle, but rather a "super supply shortage cycle" caused by an unprecedented surge in demand driven by AI computing, extreme complexity in manufacturing and packaging processes for HBM, and insufficient elasticity in the supply of general DRAM/NAND.** GPUs are responsible for generating intelligence, HBM/DRAM is responsible for high-speed data feeding, enterprise-level NAND/eSSD handles hot data and caching, while HDD is responsible for the long-term retention of massive cold/warm data. Therefore, Goldman Sachs believes that the AI arms race led by cloud computing giants is transforming memory chips from cyclical products into scarce strategic assets. The price increase of DRAM/NAND in 2026 is not the end, but may be the preliminary stage of a super cycle**SK Hynix Becomes a Key Gateway in the AI Arms Race** According to recent media reports, six informed sources have stated that the company's (SK Hynix) largest customers have been proposing a series of unprecedented pricing models to this South Korean chip manufacturing giant, including investments in dedicated memory chip production lines—where the entire production line belongs to a specific major customer. Three of the informed sources indicated that **another proposal involves customers providing substantial financial support for the procurement of semiconductor manufacturing equipment, such as ASML's extreme ultraviolet (EUV) lithography machines or even more expensive high-NA lithography machines; these devices are used for printing circuits, etching, depositing thin films, CMP, and other cutting-edge chip manufacturing processes on silicon wafers, valued at billions of dollars.** However, two of the sources stated that this South Korean chip manufacturer has a very strong cash flow, **and is cautious about accepting customers' financial and investment commitments, as such transactions could tie it to a specific buyer and may further require it to supply memory chips at prices lower than market rates in exchange for longer-term, more stable revenue guarantees.** One insider remarked, “Regardless of the type of their offers, the available memory capacity is basically zero right now.” “There isn’t even a small portion of additional capacity that can be designated for a specific customer.” Another informed source mentioned that one proposal targets the first phase of a large wafer fab that SK Hynix is constructing in the Yongin chip industrial park in South Korea, which will likely focus on dynamic random-access memory (DRAM) chip products. The details of these pricing models are being reported for the first time. SK Hynix, along with its major competitors Samsung Electronics and Micron Technology, has stated that they are negotiating multi-year supply contracts with large customers but did not provide specific details of the agreements. The sources cited by the media declined to be named due to the confidential nature of the discussions. **A spokesperson for SK Hynix refused to provide any details about the contract terms with its customers but stated, “We are thoroughly reviewing various pricing methods and structural alternatives that differ from traditional long-term agreements.”** This company is the third-largest in Asia by market capitalization, following TSMC and Samsung; thanks to investors' increasing bets on the storage supercycle driven by the frenzy of AI infrastructure, its stock price has surged 154% this year, reaching record highs. As of Friday's regular trading close on the Korean stock market, Samsung Electronics' market capitalization remained above $1 trillion. Samsung Electronics reached a milestone on Wednesday, with an approximately 80% increase year-to-date, achieving a market capitalization of $1 trillion, becoming the second Asian company to reach a $1 trillion market cap after TSMC. AI training projects rely more on large-scale parallel computing, while inference, especially for long context, multi-turn dialogue, and Agentic AI workflows, requires continuous preservation of KV Cache, context states, and intermediate results; when memory/storage space is insufficient, the model has to recompute historical states, leading to decreased GPU utilization and increased token generation costsTherefore, **HBM, DDR5, LPDDR, enterprise-grade SSDs, and even HDD/data lakes are forming an "AI memory chain" from the GPU near end to the far end storage, determining the throughput, latency, concurrency capability, and unit token economics of AI systems. This is also why storage and data storage stocks such as Micron, Samsung, SK Hynix, SanDisk, and Western Digital have experienced a surge: the demand is not only concentrated on HBM but is spilling over along the AI server architecture to the entire chain of DRAM, NAND, SSD, and HDD.** Against the backdrop of this unprecedented AI infrastructure-driven so-called "super cycle of storage chips," **the two super storage chip giants headquartered in South Korea—Samsung Electronics and SK Hynix—together account for nearly 50% of the weight of the Korea Kospi Composite Index, making them the strongest engine attracting global capital and the core driving force behind the Korean stock market repeatedly hitting new highs and significantly outperforming global stock markets.** The benchmark stock index of the Korean stock market—the Kospi Composite Index—has seen a rise of over 76% so far in 2026, surpassing last year's leading global stock market performance, but unlike the entirety of 2025, the rise in 2026 has occurred in less than five months since the beginning of the year. It is currently unclear which global tech giants are proposing investment offer models to SK Hynix, but sources indicate that these large clients include, but are not limited to, major U.S. cloud computing and tech leaders such as Google (GOOGL.US), Meta (META.US), and Microsoft (MSFT.US)—these tech giants announced last week their plans to significantly increase spending on AI computing infrastructure. Meta stated in its earnings call: "We are actively investing to meet our computing infrastructure needs," adding that this includes "reaching a series of agreements throughout the supply chain to ensure the necessary AI components for future capacity." Microsoft's management also indicated during the call that overall capital expenditures are expected to rise significantly to $190 billion this year, which includes an additional $25 billion in spending due to rising costs of chips and other components. **Record-breaking Upward Cycle of Prosperity** The offers made to SK Hynix are quite unusual for the storage chip industry. **The industry has historically been known for its extreme boom-bust cycles, and these offers are prompting storage chip manufacturers to believe that the intensity of this cycle has already surpassed the record-breaking upward cycle of the storage chip super cycle in 2018 and will last longer.**\*\* SK Hynix and Samsung Electronics' management stated last month that the current shortage of memory chips will persist for a long time, as chip manufacturers need more time to build additional supply and capacity to keep up with the "exponential growth" in AI computing demand. SK Hynix noted at the time: "Due to current supply constraints, there are significant limitations in meeting all customer demands." They also added that the demand for long-term supply contracts proposed by customers to ensure supply is increasing sharply. Memory chip manufacturers have indicated in recent years that **multi-year contracts will help smooth out severe demand fluctuations and reduce the enormous investment risks in this cyclical industry; the industry often requires billions of dollars in investment to significantly expand capacity.** The unprecedented wave of AI infrastructure and the memory supercycle have pushed semiconductors into a new stage characterized by "material intensity, process control intensity, and advanced packaging technology": **the combination of three-dimensional structures on the logic side with new materials, HBM stacking and interconnect upgrades on the memory side, and CoWoS/hybrid bonding on the packaging side transform system performance into manufacturing difficulty—these three forces collectively increase the value density of key processes such as deposition/etching/CMP/advanced packaging/core measurement, and have more clearly rewritten semiconductor equipment-related demand from "cyclical fluctuations" to "structural large expansion cycles."** The most critical semiconductor equipment base for memory chip expansion **not only includes ASML lithography machines but also encompasses expensive high-end equipment required for HBM/DRAM/NAND, such as high aspect ratio (HAR) etching/deposition, CMP (chemical mechanical polishing), metrology/testing, and hybrid bonding.** At the ubiquitous semiconductor equipment giant Applied Materials (AMAT.US), in its latest technical interpretation, it pointed out that the HBM manufacturing process adds approximately 19 additional materials engineering steps compared to traditional DRAM, claiming that its most advanced semiconductor equipment covers about 75% of these steps. It also heavily released a hybrid bonding system aimed at advanced packaging/memory chip stacking, making HBM and advanced packaging manufacturing equipment a strong growth vector for the company in the medium to long term. However, executives, investors, and analysts in the chip industry generally express that **there are still a series of unresolved issues regarding how to ensure customers do not cancel transactions and how to price memory chips favorably.** Samsung stated that, unlike previous long-term agreements, its recent agreements with some customers are "binding," but did not provide further details. A source who attended internal briefings from chip manufacturers indicated that a structure discussed as part of long-term agreements between chip manufacturers and their largest customers involves a price range mechanism; this mechanism would set a floor and ceiling for annual contract pricing and effectively eliminate quarterly or seasonal price negotiations. Another structure discussed internally by chip manufacturers involves prepayments, requiring customers to pay 30% to 40% of cash upfront. However, another source indicated that \*\*chip suppliers are acting very cautiously in how they allocate scarce capacity to avoid antitrust regulatory scrutiny or being perceived as favoring specific customers"They don't want to 'bet on a single horse' in the AI race and end up backing the wrong one," \*\* said the informed source ### Related Stocks - [DRAM.US](https://longbridge.com/en/quote/DRAM.US.md) - [MU.US](https://longbridge.com/en/quote/MU.US.md) - [MUU.US](https://longbridge.com/en/quote/MUU.US.md) - [SNDK.US](https://longbridge.com/en/quote/SNDK.US.md) - [SNXX.US](https://longbridge.com/en/quote/SNXX.US.md) - [WDC.US](https://longbridge.com/en/quote/WDC.US.md) - [STX.US](https://longbridge.com/en/quote/STX.US.md) - [MSFT.US](https://longbridge.com/en/quote/MSFT.US.md) - [GOOGL.US](https://longbridge.com/en/quote/GOOGL.US.md) - [GOOG.US](https://longbridge.com/en/quote/GOOG.US.md) - [AMZN.US](https://longbridge.com/en/quote/AMZN.US.md) - [ASML.US](https://longbridge.com/en/quote/ASML.US.md) - [NVDA.US](https://longbridge.com/en/quote/NVDA.US.md) - [SSNGY.US](https://longbridge.com/en/quote/SSNGY.US.md) - [GS.US](https://longbridge.com/en/quote/GS.US.md) - [TSM.US](https://longbridge.com/en/quote/TSM.US.md) - [META.US](https://longbridge.com/en/quote/META.US.md) - [NVD.DE](https://longbridge.com/en/quote/NVD.DE.md) - [SMSN.UK](https://longbridge.com/en/quote/SMSN.UK.md) - [W4VR.SG](https://longbridge.com/en/quote/W4VR.SG.md) - [SNDKV.US](https://longbridge.com/en/quote/SNDKV.US.md) - [WDCVV.US](https://longbridge.com/en/quote/WDCVV.US.md) ## Related News & Research - [Why Is Micron Technology Stock Surging On Friday?](https://longbridge.com/en/news/285722715.md) - [Micron: The only advanced storage island in the US mainland is evolving into an infrastructure stock.](https://longbridge.com/en/news/285335340.md) - [Top Western Digital Executive Makes Major Insider Move](https://longbridge.com/en/news/285296400.md) - [Micron vs. SanDisk: Which stock is the better buy for the AI boom?](https://longbridge.com/en/news/284976473.md) - [Seagate (STX) Had a Blowout Q3. The AI Storage Bull Case Is Building](https://longbridge.com/en/news/285493235.md)