The "gap between rich and poor" in storage chips is widening—how are winners and losers really chosen?

The "gap between rich and poor" in storage chips is widening—how are winners and losers really chosen?

The wave of artificial intelligence is reshaping the valuation logic of the semiconductor industry. As AI-driven demand explodes, the storage chip market is undergoing a significant divergence.

According to ZF Trading Desk, a Morgan Stanley research report published on January 20 indicates that in the first half of 2026, DRAM (Dynamic Random Access Memory) prices continue to rise year over year, but this is not a "feast for all." The market is being rapidly divided into two camps: the "winners" benefiting from AI inference and high-performance computing demand, and the "losers" constrained by macro headwinds and cost inflation.

Morgan Stanley analyzes that the technological inflation brought by AI is intensifying cost pressure in the supply chain, widening the "gap between rich and poor" between storage chip makers and traditional hardware manufacturers. Analysts note that AI not only drives strong demand for HBM (High Bandwidth Memory) and enterprise-grade SSDs, but even leads to persistent shortages of traditional storage products like DDR4. However, this increase in upstream storage costs is evolving into daunting profit pressure for downstream PC and smartphone OEMs.

In this "K-shaped" recovery, investors should pay close attention to companies at the core of the AI supply chain. Morgan Stanley clearly lists SK Hynix, Samsung, and SanDisk as "most favored" targets, because they directly benefit from the AI-driven commodity cycle and the NAND super cycle. On the contrary, traditional PC peripherals and OEMs unable to pass on rising storage costs to consumers, such as Acer and Logitech, continue to face market pressure.

The core logic of this trend lies in AI's "crowding-out effect." As fabs and supply chains prioritize the production of AI-related chips, non-AI sectors experience squeezed capacity, resulting in tight supply across everything from high-end HBM to traditional DDR4. For investors, understanding this supply-demand mismatch is the key to positioning for the semiconductor market in 2026.

Winners: Core Assets Riding the AI Wave

In the storage chip domain, AI is unquestionably the strongest catalyst at present. Morgan Stanley's report points out that suppliers with advanced storage technology and capacity are seeing a revaluation, driven by surging demand for AI servers and inference.

SK Hynix and Samsung are top picks, with price upside potentials of 13% and 14%, respectively. This is based on their dominance in the HBM market and improvements in the overall commodity cycle. Meanwhile, the NAND market is entering a "super cycle" supported by AI inference demand. SanDisk and KIOXIA are favored for their enterprise SSD potential. Analysis shows that AI storage demands are causing NAND shortages, and even NOR Flash supply shortages may persist until 2026.

Beyond direct storage chip makers, semiconductor equipment (SPE) companies are also beneficiaries of this boom. ASML gains from increased EUV layers, with a target price increase to 1,400 euros and 25% upside potential. Japanese equipment manufacturers like Advantest and DISCO, whose products are indispensable in HBM production, are also viewed as high-growth picks.

Additionally, among Greater China tech firms, TSMC remains the undisputed core. As the dominant foundry for AI logic chips, TSMC is planning aggressive expansions in advanced packaging (CoWoS) capacity, and its 2026 capex and structural profit margins are showing robust improvement. Morgan Stanley expects that with Nvidia’s Rubin chips ramping in 2026 and Apple A20 processors adopting the N2 process, TSMC will sustain its dominance in high-end process nodes.

Losers: Struggling Amid Cost Inflation

In stark contrast to the upstream storage manufacturers' celebration is the severe challenge facing downstream hardware OEMs. Morgan Stanley analysts warn that storage cost inflation is a challenge for the entire industry, especially for consumer electronics brands lacking pricing power.

Acer is listed as one of the "least favored" stocks, with a target price of just NT$20 and a potential downside of up to 25%. The core reason is that the increase in storage component prices cannot be fully passed on to end consumers, which will directly erode OEM profit margins. Likewise, HP and Dell face similar headwinds. Even as they attempt to enter the AI PC field, rising storage costs remain a primary negative factor in the short term.

This pressure is also affecting the peripherals and RF (radio frequency) sectors. Logitech SA is expected to be impacted by storage inflation, with these costs not being offset by macroeconomic headwinds but instead exacerbating operational pressures. RF giants Qorvo and Skyworks Solutions are also viewed pessimistically due to high costs potentially suppressing downstream hardware demand, which analysis suggests will further weaken demand.

Supply-Demand Mismatch: Not Just a Shortage of High-End Chips

Importantly, the impact of AI is not confined to high-end chips; it is causing a chain reaction across the entire storage supply chain. Morgan Stanley's research highlights one key phenomenon: DDR4 shortages will persist until the second half of 2026.

This shortage is not caused by a surge in DDR4 demand itself, but rather because supply chain capacity is being occupied by high-end AI-related products such as HBM and DDR5. As fabs prioritize high-margin AI chips, traditional DDR4 and DDR3 capacity is reduced, causing tight supply and rising prices. This benefits niche memory manufacturers such as Taiwan’s Winbond Electronics and Nanya Technology, which enjoy better pricing power.

Furthermore, AI storage demand has led to increases in NAND wafer spot prices and module prices. As capital expenditure by cloud service providers (CSPs) is expected to reach $632 billion in 2026 for building AI training and inference data centers, this will consume massive storage resources. This not only pushes up enterprise storage prices, but also drives a recovery in consumer NAND pricing, benefiting controller makers like Taiwan's Phison Electronics Corp and Silicon Motion.

Long-Term Logic: The Rise of a Trillion-Dollar Market

Looking ahead, Morgan Stanley reiterates its long-term bullishness on the AI semiconductor market. Nvidia’s CEO once predicted global cloud capex would reach $1 trillion in 2028, and Morgan Stanley’s data supports this trend, projecting global semiconductor market size could hit the $1 trillion mark by 2030.

In this grand narrative, technological inflation, AI’s self-cannibalization, and technological diffusion will be the three major drivers of the market. While emerging AI models like DeepSeek demonstrate lower inference costs, this does not change the industry’s overall reliance on high-performance computing and massive storage needs.

For investors, the strategy is now very clear: embrace upstream suppliers with core technology whose capacity is filled by AI demand, and avoid hardware assemblers downstream who are constrained by cost hikes and lack the ability to pass those costs on. Storage chip prices are not just the thermometer for the industry cycle, but also the watershed dividing market winners from losers.

 

~~~~~~~~~~~~~~~~~~~~~~~~

The above brilliant content comes from ZF Trading Desk.

For more detailed analysis, including real-time interpretations and frontline research, please join [ZF Trading Desk • Annual Membership]

Risk Warning and DisclaimerThe market has risks; investments need caution. This article does not constitute personal investment advice, nor does it take into account the special investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article fit their specific circumstances. Invest accordingly at your own risk.