Memory chip giant Micron Technology announced plans on Monday to construct a second major semiconductor fabrication facility at one of its recently acquired sites from Powerchip Semiconductor Manufacturing Corp. This move underscores the explosive growth in demand for memory chips, including High Bandwidth Memory, DRAM, and NAND. Recent statements from industry leaders, such as NVIDIA's CEO emphasizing the company's willingness to purchase all available HBM supply and AMD's CEO planning an alliance with Samsung Electronics to secure large-scale HBM shipments, highlight the intense market demand. This trend is driving major memory chip manufacturers like Micron, SK Hynix, and Samsung Electronics to accelerate production expansion efforts.
In a recent statement, Micron indicated that the new facility will significantly boost its production capacity for advanced data-center-grade DRAM products, particularly High Bandwidth Memory systems, to meet the rapidly growing computational needs of artificial intelligence applications. The company also confirmed it has completed the acquisition of Powerchip’s Tongluo P5 wafer fab site, with the new facility expected to be comparable in scale to its existing plant in Miaoli. Construction is projected to commence before the end of Micron’s 2026 fiscal year.
This expansion initiative reflects Micron’s strategy to prepare for rising demand for HBM and server-grade DRAM products while supply constraints in the AI-driven memory market persist. The goal is to swiftly convert new production capacity into volume shipments and improved average selling prices and profitability.
The push for increased memory chip production comes as a "memory super-cycle" sweeps across global markets, fueled by seemingly insatiable demand from AI data centers for HBM, server-grade DDR5 memory, and high-performance enterprise SSDs and HDDs. According to strategists from financial institutions such as Nomura, Citigroup, and Morgan Stanley, this strong demand cycle is expected to continue into early 2028.
Media reports citing informed sources indicate that AMD CEO Lisa Su is scheduled to meet with Samsung Electronics Chairman Lee Jae-yong in South Korea this week. The discussions are expected to focus on strengthening cooperation in securing HBM supply for AI chip components. For AMD, which aims to gain market share in the trillion-dollar AI chip sector dominated by NVIDIA, such partnerships may not immediately challenge NVIDIA’s CUDA ecosystem but can provide crucial supply chain stability amid current shortages.
Whether in large-scale AI computing clusters like Google’s TPU systems or NVIDIA GPU-based setups, HBM memory systems are essential components integrated with AI chips. Beyond HBM, companies like Google and OpenAI are also procuring server-grade DDR5 memory and enterprise SSDs in large quantities for new and expanding AI data centers.
From a cross-sector perspective spanning semiconductors and AI infrastructure, memory chips are uniquely positioned to benefit from the AI boom. They serve critical roles in both AI training and inference processes and are platform-agnostic, acting as universal enablers across different architectures. As AI applications evolve from training-heavy to inference, agent-based, long-context, and retrieval-augmented models, system requirements for capacity, bandwidth, power efficiency, and data persistence will only intensify.
Unlike companies such as Seagate and Western Digital, which focus on high-capacity HDDs, or SanDisk specializing in enterprise SSDs, leading memory manufacturers like Samsung, SK Hynix, and Micron hold key positions across multiple high-demand segments—HBM, server DRAM, and high-end data center SSDs. These players are positioned to capture significant benefits from the AI infrastructure boom.
Multiple international media reports indicate that Samsung, the largest player in the memory industry, has raised prices for DRAM products by over 100%. According to Korean media outlet Electronic Times, Samsung concluded negotiations with major clients, including Apple, last month, resulting in an approximate 100% increase in average DRAM contract prices for servers, PCs, and mobile devices compared to the previous quarter—effectively doubling prices from the fourth quarter of last year, with some customers facing increases exceeding 100%. Industry sources suggest that negotiations have largely concluded, with some overseas clients having already made payments. This represents a sharp acceleration from the 70% increase discussed in January, reflecting a further 30-percentage-point rise in just one month.
Rapidly rising prices for DRAM and NAND products are reshaping long-term contract practices in the global memory industry. Growing reliance on HBM, DRAM, and enterprise SSDs within GPU and TPU systems is exacerbating supply-demand imbalances. Contract negotiation cycles have shortened from annual to quarterly agreements, and in some cases now require monthly adjustments, highlighting severe market tightness.
Despite escalating geopolitical tensions in the Middle East and rising oil prices dampening investor risk appetite, analysts at Bank of America recently issued a report stating that the global memory industry remains firmly within a super-cycle. They noted that geopolitical conflicts have had negligible impact on memory supply chains or investor optimism toward the sector.
Micron is scheduled to report its second-quarter fiscal 2026 earnings after the U.S. market closes on March 18. Wall Street consensus expectations are notably high, with revenue projected around $191.5 billion and non-GAAP adjusted earnings per share estimated between $8.60 and $8.66. Compared to management’s prior guidance for the first quarter—which suggested a revenue midpoint of $187 billion and adjusted EPS midpoint of $8.42—current market expectations imply that Micron is likely to meet or exceed the upper end of its outlook.
Year-over-year, Micron’s actual performance for the second quarter of fiscal 2025 was revenue of $80.53 billion, non-GAAP EPS of $1.56, and GAAP EPS of $1.41. Based on current consensus estimates, this would represent a projected revenue increase of approximately 137.8% and a non-GAAP EPS surge of about 445% to 450%. Analysts widely anticipate not just moderate growth but a blockbuster earnings report, potentially exceeding consensus estimates, driven by relentless demand from AI data center construction.
BNP Paribas recently projected that DRAM contract prices will rise by about 90% quarter-over-quarter in the first calendar quarter of 2026, while NAND prices—traditionally more stable—are expected to increase by approximately 55%, continuing the upward trajectory observed since the second half of 2025. BNP Paribas analysts have set a 12-month price target of $500 for Micron stock. As of Friday’s close, Micron shares ended at $426.130.
Comments