HBM demand continues to explode! Micron Technology (MU.US) accelerates expansion by building another chip factory.

The American storage chip manufacturing giant Micron Technology, Inc. (MU.US) announced on Monday that it plans to build a second large chip manufacturing facility at one of the flagship wafer fabrication sites acquired from one of the leaders in Taiwan chip manufacturing, Powerchip (PSMC.US). From NVIDIA Corporation CEO Huang Renxun recently calling on global storage chip manufacturers to increase HBM production capacity as much as NVIDIA will buy, to AMD CEO Su Zi Feng planning to ally with Samsung Electronics to lock in HBM supply on a large scale, it is evident that the demand for storage chip products, including HBM, DRAM, and NAND, is experiencing exponential growth, driving Micron, SK Hynix, and Samsung Electronics, the three major original storage chip manufacturers, to accelerate production. Micron stated in a recent statement that the new chip manufacturing facility would help significantly expand the supply of its most advanced data center-level DRAM products, which mainly include high-bandwidth storage (HBM storage systems) to support the continuous surge in demand for artificial intelligence computing power. Micron also stated that it has officially completed the acquisition of Powerchip's P5 wafer fabrication plant, and the scale of the new second facility will be similar to its existing wafer fabrication plant in Miaoli. The new construction project is expected to begin before the end of Micron's fiscal year 2026. With Micron's latest ambitious production expansion plans, the storage chip giant aims to prepare in advance for the increasing demand for HBM and server-level DRAM products driven by AI, and quickly turn the new production capacity into large-scale shipments and increase ASP/actual profits. Exploding demand for storage chips As Micron accelerates the expansion of storage chip production, the global storage industry is currently experiencing a "storage super cycle" sweeping the world, with the insatiable demand from artificial intelligence data centers for HBM, server-level DDR5 storage, and enterprise-level high-performance SSD/HDD. According to strategists from Wall Street financial institutions such as Nomura, Citigroup, and Morgan Stanley, this wave of super demand for storage chips is expected to continue until early 2028. Media reports citing sources familiar with the matter revealed that NVIDIA Corporation's strongest competitor, the leading company in high-performance chips for PCs and data centers, AMD CEO Su Zi Feng, is scheduled to meet with Samsung Electronics Chairman Lee Jae-yong in South Korea this week. The two parties will focus on actively collaborating to ensure a stable supply of high-bandwidth storage (HBM storage systems) for artificial intelligence chip components. For AMD, which is striving to continually erode NVIDIA Corporation's market share of up to 90% in the trillion-dollar AI chip market, this cooperation may not immediately undermine NVIDIA Corporation's CUDA ecosystem, but it can bring a more practical advantage in the current situation where storage chip demand far exceeds supply: supply chain certainty. Whether it is Alphabet Inc. Class C's immensely huge TPU AI computing cluster or the massive NVIDIA Corporation AI GPU computing cluster, both rely on HBM storage systems that need to be fully integrated with AI chips. Besides HBM, Alphabet Inc. Class C and technology giants like OpenAI are accelerating the construction or expansion of AI data centers, which also require large-scale purchases of server-level DDR5 storage, enterprise-level high-performance SSD/HDD, and other storage solutions. From the perspective of semiconductor and AI data center infrastructure, storage chips are perfectly positioned to ride the AI wave because they can handle both training expansion and inference expansion. They are also a "universal toll booth" that spans platforms, architectures, and ecosystems. As the AI era transitions from training dominance to inference, agent models, long-context models, and enhanced retrieval models, the demand for capacity, bandwidth, power efficiency, and data persistence will only increase. Unlike Seagate and Western Digital Corporation, which focus on dominating large capacity nearline HDDs, SanDisk focuses on high-performance eSSDs. Samsung Electronics, SK Hynix, and Micron, the three major storage chip manufacturers, are strategically positioned in multiple core storage areas: HBM, server DRAM (including DDR5/LPDDR5X), and high-end data center enterprise-level SSDs (eSSDs). They are the most direct beneficiaries in the "AI memory + storage stack" and are reaping the rewards of AI infrastructure. Several overseas media reports have indicated that Samsung, the largest player in the storage industry, has raised the prices of its dynamic random access memory (DRAM series storage products) by over 100%. According to Korea Electronics News, Samsung Electronics completed final negotiations with its largest customers, including Apple Inc., on first-quarter DRAM supply prices last month. The average price of server, PC, and mobile universal DRAM has increased by approximately 100% from the previous quarter and doubled from the fourth quarter of last year, with some customers and products even seeing increases of over 100%. The report cited insiders in the industry revealing that the negotiations have been fully concluded, and some overseas customers have completed payments. This increase is higher than the 70% level negotiated in January and has expanded by approximately 30 percentage points in just one month. The rapid rise in prices of DRAM/NAND series storage products is reshaping the long-term contract practices of the global storage industry, especially as the GPU/TPU ecosystem increasingly relies on HBM, DRAM, and enterprise-level SSDs. The supply negotiation cycle has compressed from traditional annual contracts to quarterly contracts and now requires adjustments on a monthly basis, reflecting the severity of the imbalance between supply and demand in the storage chip market. Market expectations for Micron's explosive performance As the war between the United States/Israel and Iran escalates and spreads to multiple countries in the Middle East, igniting a new round of political superstorms that sweep the global economy with the GEO Group Inc, investors have become increasingly risk-averse against the backdrop of soaring oil and gas prices. Their serious concerns about the global economy, which is still in a fragile recovery process and may face "stagflation" due to out-of-control surges in energy prices, have led to significant turmoil in global stock, bond, and cryptocurrency markets recently. However, a recent research report from the analyst team at Bank of America Corp on Wall Street indicates that the global storage industry, with storage chips at its core, is still in the midst of a "storage super cycle," and the impact of the political conflict in the Middle East by GEO Group Inc on the storage supply chain and the bullish sentiment of fund managers towards the storage sector is almost negligible. Micron is set to release its performance for the second quarter of fiscal year 2026 after the US stock market closes on March 18. The mainstream consensus on Wall Street for this earnings report is very high: revenue is expected to be around $19.15 billion, and non-GAAP adjusted EPS is expected to be in the range of $8.60 to $8.66. Based on the guidance provided by the company's management for the previous quarter (Q1), with revenue at a median of $18.7 billion and adjusted EPS at a median of $8.42, it means the market is already betting heavily that Micron will at least hit the upper end of the guidance, and some analysts even expect it to slightly exceed the outlook given by Micron's management. In comparison to the same period last year, Micron's actual performance in Q2 of fiscal year 2025 was: revenue of $8.053 billion, non-GAAP adjusted EPS of $1.56, and GAAP EPS of $1.41. If calculated based on the current Wall Street consensus expectations, this means that Micron's revenue for the second quarter is expected to increase by about 137.8% year-on-year, and non-GAAP EPS is expected to increase by 445% to 450%. In other words, the market is expecting not just "moderate growth," but an explosive growth in the earnings report, and analysts widely believe that with the almost never-ending storage demand driven by the frenzy of AI data center construction, Micron is likely to significantly exceed the consensus expectations. BNP Paribas recently released a research report stating that contract prices for DRAM in the first quarter of 2026 are expected to increase by 90% compared to the previous quarter, while NAND, known for its stable price curve, is expected to rise significantly by 55%, with the price increase trend continuing into the second quarter since the second half of 2025. The BNP Paribas analyst team even set a target price of up to $500 for Micron within the next 12 months. At the close of the US stock market on Friday, Micron's stock price was at $426.13.