Global smartphone shipments fell 20% year-on-year in the third quarter of 2026, their lowest 3Q level since 2012. The source explicitly states the cause: persistent DRAM and NAND shortages inflated production costs, constrained supply, and weakened demand. Samsung led with a 23% market share, while Apple reached a record third-quarter share of 22%, highlighting a widening divide between premium and price-sensitive segments.
The source states that rising complexity in AI chip manufacturing is creating new opportunities for semiconductor equipment suppliers. Applied Materials says rising transistor counts, denser interconnects, and more advanced memory architectures are expanding demand across logic, DRAM, and high-bandwidth memory (HBM) manufacturing. No specific figures, dates, or price movements are given; this is background industry information.
logicDRAMhigh-bandwidth memory (HBM)Applied MaterialsDigiTimes Asia
Memory chipmakers have reportedly become the fastest-growing source of demand for ASML, and the two largest Korean memory makers have reportedly accepted higher replacement-part prices from the Dutch company. ASML is preparing to charge more for its most critical tools. Attention now turns to TSMC, ASML's largest customer, which according to earlier reports was an opponent of the price increases; whether TSMC accepts remains an open question.
ASML replacement parts for lithography toolsASML most critical toolsASML, TSMCDigiTimes AsiaPrice Movement
Global DRAM demand for humanoid robots is projected to increase more than 20-fold by 2030 as robot deployments expand and memory capacity per unit doubles, potentially opening a new growth market for Samsung Electronics and SK Hynix beyond the high-bandwidth memory (HBM) used in AI data centers. This is a demand forecast, not an achieved result.
Nvidia's next-generation Rubin AI accelerators are expected to rely predominantly on 12-layer HBM4 high-bandwidth memory, despite reports that Nvidia has asked Samsung Electronics and SK hynix to increase supplies of 8-layer products. The source does not state the cause of the differing assessments; it notes that the differing assessments suggest changes in memory configurations do not necessarily signal weaker HBM demand, and that manufacturing constraints and supply shortages are expected to support memory prices in 2027. The event remains an expectation/report and the final configuration is not confirmed.
12-layer HBM48-layer HBM4Nvidia, Samsung Electronics and 3 totalDigiTimes Asia
Fujifilm is expanding semiconductor materials production in Arizona (US); the source states the cause is rising global chip demand, especially for artificial intelligence, advanced logic, and memory. The move aims to strengthen supply for makers in the US and beyond, while supporting new packaging technologies and faster commercialization worldwide. The source gives no specific scale, investment amount, or timeline, and the expansion remains a planned capacity increase.
Topco Scientific (TSC) is evaluating a new base in Boise, Idaho, and plans to strengthen its local technology, engineering, and supply chain capabilities to pursue memory and advanced-process opportunities. The company already operates service bases in Phoenix, Arizona, and Dallas, Texas. The source links this plan to the US entering a fresh wave of chip manufacturing expansion and Micron ramping up advanced memory investment. The event remains at the evaluation/planning stage and is not a confirmed implementation outcome.
According to Reuters, citing sources, SK hynix is reportedly seeking U.S. approval to expand chip production in China via its U.S. subsidiary Solidigm, which aims to install additional equipment in unused space at the NAND flash facility in Dalian, China — specifically Mod B, a fabrication space completed in 2023 whose equipment installation was left unfinished due to the memory market downturn — as a condition for investing in a U.S. facility, with the expanded operations helping fund a new NAND manufacturing plant in the U.S.; Solidigm is also reportedly considering an initial public offering as early as next year to raise additional capital. However, SK hynix told Reuters that equipment installation at Dalian's Mod B facility is proceeding in consultation with and under authorization from the U.S. government, and that it is not seeking to expand capacity at Mod B; Solidigm declined to comment. As background: in September 2025, the U.S. Commerce Department revoked special authorizations that had allowed SK hynix and Samsung to receive U.S. chipmaking equipment at their Chinese facilities without individual export licenses, saying it would continue allowing shipments needed to maintain existing operations but did not intend to approve exports for capacity expansion or technology upgrades; in December 2025, Washington granted annual approval for U.S. chipmaking equipment shipments to Samsung and SK hynix's Chinese facilities for 2026, per Yonhap News, though Reuters reports the authorization for the Dalian plant limits equipment quantities and approval for additional U.S.-made equipment and technology in 2027 has yet to be issued. Per The Bell, access to advanced etching equipment is particularly important as NAND makers increase stacked layers, and using older tools can require additional processing steps, reducing productivity and raising manufacturing costs; per Reuters, the Dalian NAND facility uses equipment from major U.S. suppliers including Lam Research, Applied Materials, and KLA.
NAND flashSK hynix, Solidigm and 3 totalTrendForcePolicy/Regulation
Researchers at ETH Zürich and New York University published a technical paper titled “Understanding the Power Consumption of Modern High Bandwidth Memory: Experimental Analysis and Modeling Using Real HBM2 DRAM Chips.” According to an abstract excerpt, they perform the first comprehensive experimental characterization of power consumed by real HBM2 DRAM chips, and their extensive characterization yields 1) the empirical... (excerpt truncated). The event is an academic research publication with no stated cause and no direct price or supply implications.
According to a report published by The Register on Wednesday, 14 October 2026 at 16:00 UTC, Gigabyte recently announced new motherboards that use DDR4 RAM, a standard that launched in 2014. For AM4 processors, Gigabyte offers the new B450M H ARGB with two DDR4 slots and the B550 X-Eagle with four slots, while Intel-platform buyers can get the B760M Aorus Elite. The article states this is partly due to the current price rises and component shortages induced by large-scale, cloud-based LLMs. In the People's Republic of China, several vendors sell new DDR3 motherboards for circa 15-year-old server CPUs, such as Huananzhi's X79-4D or X79 Supreme, or the Machinist X79 V2.82H, all using Intel's LGA-2011 socket, which launched in 2011. The DDR5 standard was agreed in 2020, with mainstream PC adoption following in 2021. The article also argues that Linux lets older hardware remain in service, making second-hand kit more usable than ever.
SK hynix has detailed plans for an in-house core named a Hynix Transaction Processing Unit (H-TPU), intended to reduce reliance on power-hungry CPU cores. The plan was presented by Koo In-jae, the company's VP of SoC design, at ISOCC 2026 (the International SoC Design Conference), and remains a plan rather than a shipping product: H-TPU has yet to be proven in an actual enterprise SSD controller product, and no date has been given. The H-TPU uses a four-stage, in-order pipeline, is placed in clusters to handle repetitive tasks such as read/write command processing, metadata management and buffer control, with more than 100 H-TPUs in total; management remains with the traditional Arm CPU, making it an offload rather than a replacement. The cores are claimed to draw only about 2.5% of a general-purpose CPU core's power, with around 4% of its area (the CPU core used for comparison is not named). The reported cause is that SSD controller CPUs are at their limit on power and area, and PCIe bandwidth doubles every generation while controller CPU clocks and control capability reportedly cannot keep up. SK hynix reportedly plans to apply H-TPU across its NAND products, including next-gen enterprise SSD controllers, and to high-bandwidth flash (HBF), which the outlet said is due next year; the H-TPU is reportedly already being used in 'Storage Next', Nvidia's name for the joint AIN P project with SK hynix.
IN2FAB offers a Migration Feasibility Study for analog and mixed-signal IP, applying migration technology to a representative portion of IP to turn assumptions into engineering evaluations, helping companies decide between full migration or redesign by assessing reuse, showing which devices map, where parameter or device-size limits arise, which layout structures can be preserved, where engineering intervention may be needed, identifying where automation can make the greatest contribution, assisting in estimating effort, assigning resources, developing schedules, defining verification requirements, guiding delivery approach (internal use, engaging IN2FAB for engineering services, or combining both), and following the process: start with design, test migration, measure effort, then decide how to proceed. The source does not state the cause.
ASML's Q3 2026 financial results show shipments to China surged to 23% of system sales, up from 14% in the previous quarter, becoming one of its strongest areas. Chinese semiconductor manufacturers are rushing to stockpile more tools. Taiwan remains the largest buyer at 32%, followed by South Korea at 31%. The United States is reportedly pressuring the Dutch government to completely halt tool exports to China. CXMT and YMTC have secured sufficient DUV lithography tools from ASML to support their growth plans for the next three years.
shipments to Chinasystem salesASML, Chinese semiconductor manufacturers and 4 totalTechPowerUpInventory Shift
Related manufacturers
ASML, Chinese semiconductor manufacturers, CXMT, YMTC
Rambus launches a new PMIC designed for LPDDR6 SOCAMM2 modules to enhance power efficiency in AI infrastructure. This technical advancement supports the growing demand for low-power, high-performance server memory solutions.
Samsung Electronics unveiled its next-generation memory strategy at the Open Compute Project (OCP) Global Summit, an IT infrastructure event held in California on Oct. 12. Shim Ho-joon, vice president of the AE Team in Samsung's DS Division Memory Business, outlined three strategic priorities: speed, scalability and persistence. The stated reason in the source is that as AI models grow, the amount of data they must process is increasing, and Shim said memory capacity expansion and data storage technologies are becoming as important as memory bandwidth. Samsung plans to improve data processing speeds and computational efficiency through eighth-generation high-bandwidth memory (HBM5), custom HBM, zHBM and processing-in-memory (PIM) technologies; to expand AI system memory capacity and improve data management efficiency using CMM-D and Compute Express Link (CXL)-based technologies; and to increase AI data storage capacity and improve management efficiency with next-generation solid-state drives (SSDs), quad-level cell (QLC) technology and flexible data placement (FDP). These are stated plans, a strategy announcement rather than achieved results. Samsung also showcased the results of its collaboration with Nvidia, displaying a Groq wafer — Nvidia's language processing unit (LPU) — and the Groq 3 LPX system; its foundry business manufactures Groq chips using its 4-nanometer process. Samsung also jointly demonstrated a CXL-based memory expansion solution with AMD, and the two companies are collaborating with global customers on designs from the early stages of product development and expanding technology validation.
HBM5custom HBMzHBMSamsung Electronics, Nvidia and 4 totalThe Elec
The AI boom has prompted memory manufacturers to expand HBM production while cutting legacy DRAM capacity, tightening DDR4 supply and causing DDR4 contract prices to rise faster than DDR5, reversing the usual pricing pattern in the memory market.
Taiwan's export surge is increasingly tied to high-value AI servers and pricier memory chips; as exports climb, the outlook underscores both global demand for artificial intelligence hardware and the risks of relying heavily on one fast-growing sector. The source does not state specific figures or causes.
ChangXin Memory Technologies (CXMT) plans to mass-produce next-generation 4F² DRAM by the end of 2026, potentially ahead of South Korean rivals Samsung Electronics and SK Hynix. The announcement has fuelled speculation that China's memory technology is approaching or surpassing South Korean capabilities, while positioning CXMT for an early entry into the emerging 3D DRAM market. This remains a plan; mass production has not yet been achieved.
Qualcomm said it is pushing to cut edge AI's energy cost as phones, cars, and other devices face tighter power and cooling limits than data centers. At the opening of Applied Materials' EPIC Center, the company said its memory-and-logic stacking work could sharply reduce energy use, although thermal, materials, and manufacturing challenges remain globally. The source does not state a timeline for commercialization of this technology.
On October 14, 2026, TrendForce reported from the OCP Global Summit 2026 that Arm said the CPU market is expected to surge from $52 billion in 2026 to $300 billion by 2030 due to agentic AI redefining data center architectures; Arm’s AGI CPU debuted in March 2026.