The unprecedented global wave of artificial intelligence deployment is poised to fuel chip demand—particularly for AI chips and memory chips—driving explosive growth trajectories over the next decade. Divya Mathur, an emerging markets equity portfolio manager at ClearBridge Investments, whose fund performance outpaces the vast majority of global peers, asserts that memory chip stocks and shares of high-end storage products represent the most compelling investment area in global equity markets. This veteran fund manager's proclamation that memory chips are the optimal investment choice coincides with Wall Street giants Morgan Stanley, Nomura, and Bank of America heralding the full arrival of an AI-driven "storage chip super-cycle." They suggest the intensity and duration of this cycle could far surpass the "cloud computing-era driven storage super bull market" of 2018. Extremely bullish global sentiment towards memory chips has propelled South Korea's benchmark Kospi index to a staggering 76% surge in 2025, arguably the wildest stock market performance globally for the year. This was primarily fueled by explosive gains from the two leading global memory chip makers, SK Hynix and Samsung Electronics, which together hold over a 30% weighting in the index and contributed nearly half of its total gains. As the sole major U.S.-based memory chip manufacturer, Micron Technology (MU.US) also demonstrated powerful momentum, its stock soaring 240% throughout 2025 and already climbing 18% year-to-date in 2026. Furthermore, the three leading storage product makers—Western Digital (WDC.US), Seagate Technology PLC (STX.US), and SanDisk (SNDK.US)—have also posted exceptionally strong gains in 2026, following full-year 2025 increases that each exceeded 200%. Demand for DRAM/NAND memory chips remains robust, with prices for product series like DDR4/DDR5 and enterprise-grade SSDs exhibiting rapid expansion. This is primarily driven by the AI computing tidal wave elevating the demand for memory chips and their critical importance to AI training/inference systems to unprecedented levels. Global demand for AI computing power continues its exponential growth trend, far outstripping the available supply, a dynamic clearly visible in the exceptionally strong quarterly results just announced by the "world's chip king," Taiwan Semiconductor Manufacturing (TSM.US). TSMC's fourth-quarter gross margin broke through 60% for the first time, with net profit significantly exceeding expectations. The company forecasts nearly 30% revenue growth for 2026 and substantially raised its capital expenditure guidance to $52-$56 billion, with both core forecasts far surpassing market expectations. Additionally, management sharply raised the compound annual growth rate expectation for its AI-related chip foundry business from the "mid-40% range" to the "mid-to-high 50% range." The extraordinarily strong results and future guidance from the world's largest chipmaker triggered a broad rally in U.S. chip stocks on Thursday, with memory chips and semiconductor equipment shares leading the gains. The ClearBridge SMASh Series Emerging Markets Equity Fund, managed by Divya Mathur, has outperformed 97% of its peer funds over the past year according to compiled institutional data. The fund has placed significant bets on memory chip giants Samsung Electronics, whose stock doubled in 2025, and SK Hynix, whose stock nearly quadrupled during the same period. Mathur anticipates this bull trend will persist long-term, stating that the AI wave will permanently reshape this long-perceived cyclical and commoditized chip industry. "Since the semiconductor era began, the memory chip industry was never built for the data storage needs of the AI field—but in the past year or so, we have encountered this new growth driver," said Mathur, a senior portfolio manager at the Edinburgh-based investment firm and a specialist fund manager within global asset management giant Franklin Templeton, with whom he co-manages the approximately $1.4 billion emerging markets equity fund. He has held positions in these two memory chip stocks since 2015 and expresses growing, long-term conviction in their prospects. As illustrated, South Korean memory chip manufacturers' stock prices are accelerating upwards, with strong momentum expected to continue throughout 2026. South Korea's stock market experienced an unparalleled "renaissance-style rally" in 2025, arguably a historic event in global financial markets, achieving its strongest gain in 25 years. Samsung and SK Hynix contributed nearly half of the index's rise, while Korean nuclear energy firms were also major contributors. The bullish narratives for both memory and nuclear power jointly benefit from the massive computing power demand generated by AI training/inference systems. More significantly, Wall Street analysts believe the Korean market still has substantial upside potential. Top-tier brokerages including Citigroup, J.P. Morgan, and Nomura predict the benchmark index will rise at least another 20% next year. Analysts added that additional government stimulus measures and the strong earnings growth trend driven by the "memory chip super-cycle" should continue to support this "bullish upward trajectory." 2025 was a record-breaking, exceptionally glorious year for SK Hynix, the world's largest supplier of HBM storage systems, as it remained the primary HBM supplier for Nvidia's (NVDA.US) AI GPU compute clusters and Google's (GOOGL.US) TPU AI compute clusters. Meanwhile, Samsung Electronics saw its profit surge to a record level in the quarter ending December, largely due to soaring overall memory chip demand and selling prices driven by the massive infrastructure needs from the AI data center construction and expansion frenzy. Even as these memory chip leaders enter an unprecedented bull market trajectory and memory chip product price increases reach the 50%+ level, Mathur contends the market continues to underestimate the intensity of demand driven by AI infrastructure build-out. "Furthermore, based on information I have, some U.S. tech clients indicate the storage industry is only in the second year of a decade-long super upgrade cycle," he stated in a media interview. Perhaps more importantly, compared to their U.S. chip stock peers, Asian AI-related chip companies currently sport relatively reasonable valuations. Latest statistics show Samsung and SK Hynix trade at forward P/E ratios of just 9.3x and 7x respectively, compared to approximately 26x for the Philadelphia Semiconductor Index, the U.S. semiconductor benchmark. Mathur added that even within the memory chip sector, careful selection of stocks with strong fundamentals remains crucial. "The benefit of holding Samsung and SK Hynix shares is that it can compensate for potential execution missteps or some delayed returns you might face," he said. "The last thing you want in investing is to identify a sector but choose the wrong company—the one that fails to benefit from the trend." Wall Street's bullish sentiment towards memory chips is reaching fever-pitch levels. Whether it's Google's massive TPU AI compute clusters or vast arrays of Nvidia AI GPU clusters, all critically depend on integrated HBM storage systems paired with AI chips. Furthermore, tech giants accelerating the construction or expansion of AI data centers must purchase server-grade DDR5 memory and enterprise-class high-performance SSDs/HDDs on a massive scale. Samsung, SK Hynix, and Micron Technology are strategically positioned across these three core storage domains: HBM, server DRAM (including DDR5/LPDDR5X), and high-end data center SSDs, making them the most direct beneficiaries within the "AI memory + storage stack" and prime recipients of the AI infrastructure "super红利" (super dividend). Globally, the robust demand for storage components essential to AI training/inference systems from large-scale AI data centers under intense construction is exceeding supply, fully benefiting memory chip companies like Micron. Concurrently, a sharp shortage has emerged for lower-performance memory products used in consumer electronics like PCs and smartphones, collectively pushing DRAM and NAND product prices into a frenzy. This is largely attributable to the storage industry shifting capacity towards more advanced manufacturing technologies for AI data centers. The three dominant memory chip manufacturers—SK Hynix, Samsung, and Micron—are concentrating most of their capacity on HBM systems. These products require significantly more complex advanced process capacity, manufacturing, and packaging/testing compared to DDR series or HDD/SSD memory chips. Consequently, the migration of capacity to HBM by these leaders has substantially contributed to supply shortages for other storage drives. Moreover, shares of enterprise data storage product giants have also shown powerful gains. Seagate, SanDisk, and Western Digital each saw their stock prices rise over 200% in 2025, with enterprise SSD leader SanDisk achieving a staggering 500% increase. These memory chip and product line leaders have significantly outperformed both the broader U.S. market and global equity indices. The core logic behind the strong stock performance of these three storage product giants is that the fervent construction of AI data centers not only drives surging HBM demand but also necessitates simultaneous expansion across the data center's three-tier storage stack (hot-tier NVMe SSD, warm-tier/nearline HDD, cold-tier object & backup). This, combined with long-standing supply discipline among HDD oligopolies, a recovering NAND cycle, and multi-year volume lock-ins by cloud providers, has catapulted the volume, pricing, and order visibility for these three companies. A recent memory chip industry tracking research report from Japanese financial giant Nomura indicates that DRAM/NAND memory chip price ascent curves are steepening. This is driven by strong resonance from surging demand for enterprise-grade high-performance server DRAM, HBM storage systems, and data center high-performance SSDs, all fueled by the faster-than-expected acceleration of global AI data center construction. Nomura's analysts judge that this "memory chip super-cycle," which began in the second half of 2025, will last at least until 2027, with meaningful new supply additions unlikely to emerge before early 2028. Nomura advises investors to maintain an overweight position in memory leaders in 2026, focusing on the "price-profit-valuation" triple play as the main investment theme, rather than viewing memory solely through the HBM lens. The firm expects the three major memory chip companies to achieve record-high profits. Nomura's latest assessment reiterates that this "memory chip super-cycle" will persist at least until 2027, explicitly stating in its report that "meaningful supply increases [are expected] earliest in 2028." The report repeatedly emphasizes that expanding capacity for memory chips and HDD/SSD storage components is not something that can be done overnight, involving complex timelines for greenfield/brownfield projects and customized semiconductor equipment upgrade plans. Using SK Hynix as an example, Nomura precisely identifies capacity constraints related to long lead times for cleanroom/wafer capacity, significant delays in yield and cycle times due to advanced process upgrades, and limitations on building/upgrading overseas chip factories. These factors lead to the conclusion that supply increases will be slower and the supply deficit will last longer. Analysts at Citigroup, in their latest outlook, present an even more aggressive bullish stance than Nomura. They believe that, driven by the proliferation of AI Agents (AI intelligent agent workflows) and surging memory demands for AI CPUs, memory chip prices will experience runaway increases in 2026. Citi's analysts have violently raised their 2026 average selling price (ASP) growth forecast for DRAM from 53% to 88%, and for NAND from 44% to 74%. They project that server DRAM ASP will skyrocket 144% year-over-year in 2026 (up from a previous forecast of +91%), driven by both AI training and inference demand. For example, using the mainstream 64GB DDR5 RDIMM product, Citi predicts its price will reach $620 in Q1 2026, a 38% sequential increase, far exceeding the previous forecast of $518. In the NAND sector, Citi's forecasts are equally aggressive, raising the 2026 ASP growth expectation from +44% to +74%; within this, enterprise SSD ASP is expected to increase 87% year-over-year. In Citi's view, the memory chip market is entering an extremely intense seller's market, with pricing power fully concentrated in the hands of memory giants like Samsung, SK Hynix, Micron, and SanDisk.
Comments