SEMI forecasts that by 2026, investment in 300mm storage‑device manufacturing will surpass US$50 billion for the first time, as the industry undergoes structural transformation driven by AI.

The global memory industry is currently experiencing an unprecedented “supercycle.” According to the latest 300mm Fab Outlook Report released by the Semiconductor Equipment and Materials International (SEMI) on June 30, equipment investment in the memory segment of 300mm fabs worldwide is expected to surpass US$50 billion for the first time in 2026, rising 29% to US$52 billion, with a further 11% increase projected to reach US$57 billion in 2027. -1 -。 SEMI President and CEO Ajit Manocha stated, “Strong demand for HBM and other advanced memory technologies is reshaping investment priorities across the entire semiconductor supply chain.”

The underlying driving force behind this round of investment boom is The continuous expansion of AI infrastructure, data centers, and next-generation computing systems is underway. Yet beneath this apparent prosperity, the memory industry is undergoing multifaceted, deep‑seated structural transformations—changes that not only shape the industry’s trajectory but also exert far‑reaching implications across the entire electronic components value chain.

I. Surging Investment and Capacity Bottlenecks: Structural Contradictions Behind Massive Capital Flows

Capital expenditure continues to rise, but the growth in effective capacity remains moderate. According to a SEMI report, from 2024 to 2029, global equipment investment in the 300mm wafer fab storage sector is expected to grow at a compound annual growth rate of 19%. Global 300mm storage capacity is projected to reach 4.1 million wafers per month in 2026 and 4.2 million wafers per month in 2027. However, SEMI also notes that, due to technology transitions and process complexity—including the shift to advanced-node DRAM, HBM, and higher‑layer NAND— Growth in effective capacity remains moderate. 。

This means that massive investments have not yet been swiftly translated into an adequate supply. For instance, Taking HBM as an example, Feng Li, President of SEMI China, previously revealed that the HBM market is expected to grow by 58% to US$54.6 billion by 2026, accounting for nearly 40% of the DRAM market. However, despite the three major manufacturers having allocated 70% of their new production capacity to HBM, The capacity gap remains as high as 50% to 60% 。 SK Hynix has already sold out its entire HBM production capacity for 2026.

This substantial gap between investment scale and effective production capacity means that, in the advanced‑node segment, the economies of scale associated with capital expenditures are diminishing—technological complexity and process‑technology barriers have become more critical constraints than funding. As a result, upstream equipment and materials suppliers will enjoy sustained, robust demand; however, downstream component buyers must recognize that the pace of capacity expansion lags far behind the rate of demand growth.

II. Price Differentiation and Market Restructuring: The “Parallel Worlds” of AI and Non-AI

Memory prices are at historic highs, but trends across different market segments are diverging sharply.

Prices for AI-related storage products remain firm. A report released by Morgan Stanley on July 2 indicates that global NAND supply will face a 15% shortfall in 2026, with the gap projected to narrow to 9% in 2027. NAND prices for enterprise‑grade SSDs rose approximately 30% quarter over quarter in the third quarter, while consumer‑grade NAND prices increased only modestly. In the DRAM market, server memory prices are expected to climb about 20% sequentially, with traditional DDR3 and DDR4 seeing even sharper price gains—ranging from 30% to 40%—due to persistently tight supply and rising AI‑related demand.

However, consumer demand has already shown signs of fatigue, with growth rates clearly moderating. According to the latest memory price survey released by TrendForce on July 3, the overall DRAM market will remain in severe shortage through the third quarter of 2026. However, due to downward revisions in consumer‑grade demand and a high base effect, contract price increases are expected to narrow to 13%–18% quarter over quarter. NAND Flash contract prices are forecast to rise 10%–15% quarter over quarter, a significant deceleration compared with previous quarters.

Morgan Stanley explicitly advises investors to “prioritize DRAM over NAND and favor memory manufacturers over module makers.” The firm notes that AI‑driven NAND demand will grow from 205 EB in 2025 to 400 EB in 2026, and further to 609 EB by 2027, with AI’s share of total NAND demand rising from 18% in 2025 to 41% in 2027. Meanwhile, consumer‑grade NAND is facing headwinds: following price hikes in the second quarter, smartphone and PC customers have begun scaling back their orders.

The memory market is splitting into two “parallel worlds”: AI‑grade and enterprise‑grade storage are in short supply and command firm prices, while consumer‑grade storage is grappling with weakening demand and a price ceiling. This divergence will ripple through the entire electronic components supply chain: component suppliers serving AI data centers will continue to enjoy premium pricing, whereas those reliant on the consumer electronics market will face dual pressures—shrinking demand and eroding profit margins. Component manufacturers need to reassess their customer base and product positioning.

III. Accelerating Evolution of the Technological Roadmap: Advanced Packaging Emerges as the Decisive Factor

AI has not only reshaped the demand landscape for memory but has also fundamentally reoriented the trajectory of technological advancement.

The importance of advanced packaging has surpassed that of process node scaling. According to a report by China Times News Network, in the past, the memory industry primarily relied on process scaling to increase capacity and reduce costs; however, with the advent of the AI era, The importance of advanced packaging has surpassed process nodes, becoming a key determinant of both competitiveness and supply capacity. . In In the DRAM sector, the three major manufacturers will fully adopt the 1c process node by 2026 and subsequently move toward the 1d and 10a nodes—where 10a marks the official entry of storage technology into the sub‑10nm era. In the HBM space, HBM4 will be upgraded from a 12‑layer stack to a 16‑layer stack in 2026, with HBM4E incorporating hybrid bonding technology across the board.

NAND technology is also accelerating breakthroughs. On July 3, Kioxia and SanDisk jointly announced the commencement of engineering sample deliveries for their tenth-generation BiCS FLASH 3D flash memory, which employs a 332-layer cell‑stacking architecture. This chip achieves an industry‑leading TLC storage density exceeding 29 Gb/mm². - . Kioxia’s forecast From 2026 to 2028, the NAND market’s overall shipment capacity is expected to grow at a compound annual growth rate of 22%, with the data center segment posting an even higher growth rate of 46%. -。

Even more disruptive is the trend toward the convergence of storage and computing. SanDisk’s latest publicly released materials propose 3D stacking NAND flash dies beneath AI accelerators, GPUs, and other compute‑centric die architectures, while retaining HBM DRAM on the same interposer. The two work in tandem: HBM handles low‑latency, high‑frequency read/write operations, whereas NAND is tasked with high‑capacity data storage and retrieval. Meanwhile, SK Hynix has introduced its AIN product line; the AIN B variant adopts an HBF architecture similar to HBM, leveraging TSV technology to stack 16 layers of NAND, with capacities reaching up to 4 TB.

Traditionally distinct domains of “computing” and “storage” are now converging into a deeply integrated whole. Advanced packaging is no longer merely a back-end process in chip manufacturing; it has become a core technology that determines system-level performance. This shift implies unprecedented demand growth for a range of supporting components—such as packaging substrates, interposers, TSV‑process materials, and test equipment. At the same time, The blurring of the physical boundary between storage and computing will give rise to entirely new categories of components and system architectures. , bringing opportunities for restructuring across the upstream and downstream segments of the industrial chain.

IV. Supply Chain Dynamics and Geopolitics: Strategic Reconfiguration in an Era of Scarcity

The memory shortage has escalated from an industry issue to a geopolitical concern.

Terminal giants are under pressure to raise prices. Apple CEO Tim Cook described the current imbalance between storage supply and demand as a “once-in-a-century flood.” On June 25, Apple announced price hikes across its Mac, iPad, and smart‑home product lines, with individual items seeing increases of up to 3,500 yuan. Meanwhile, Apple is seeking to source chips from ChangXin Memory and Yangtze Memory Technologies to ease supply‑chain constraints and cost pressures. - Meanwhile, Micron’s Chief Commercial Officer, Sumit Sadana, subtly suggested that Apple’s past aggressive price‑cutting practices had dampened suppliers’ enthusiasm for expanding production capacity.

Antitrust enforcement and capacity expansion are unfolding in tandem. On June 25, Samsung, SK Hynix, and Micron faced a class-action lawsuit in the United States, accused of colluding to cut traditional DRAM production capacity, driving prices up by roughly 700% over four years. Almost simultaneously, the South Korean government announced an investment of 800 trillion won to build four new wafer fabs, aiming to double DRAM capacity within five years. Meanwhile, SEMI recently sent a letter to the U.S. government, warning that interventionist measures could further exacerbate supply shortages triggered by artificial intelligence.

Memory is no longer merely a component; it has become… Strategic resources Its supply security directly impacts the cost and availability of end products. For the electronic components industry, this means that diversifying the supply chain is no longer merely a business decision—it has become a matter of survival. The rise of Chinese memory‑chip companies—such as CXMT, which reported RMB 61.8 billion in revenue in 2025 and saw a 719% year‑over‑year increase in Q1 2026—is introducing new options to the global component supply chain, yet it also brings with it complex geopolitical uncertainties.

Conclusion: The “Perils” and “Opportunities” of the Supercycle

The global semiconductor market is projected to surpass US$1.51 trillion by 2026, up 90% year over year, with memory chips posting a year-over-year growth rate of 250% and exceeding US$800 billion in size. Such extraordinary growth rates are unprecedented in the history of the semiconductor industry.

However, for the electronic components industry, this round of the memory supercycle has brought not only a short-term surge in orders and profits, but also a series of profound industry‑wide insights: Technological complexity is reshaping the boundaries of investment effectiveness. The divergence between AI and non-AI markets is reshaping the rules of competition, advanced packaging and the convergence of storage and computing are blurring traditional industry boundaries, and supply-chain security is evolving from a cost‑driven concern into a strategic imperative.

Only component‑industry companies that can accurately identify structural trends, carve out a clear position amid market differentiation, and proactively plan for technological disruption will be poised to gain a competitive edge in this round of industrial reshaping. By contrast, players that rely solely on cyclical tailwinds while ignoring deeper structural changes risk being left behind when the tide turns.

The memory industry stands at an unprecedented historical juncture—presenting both challenges and opportunities.