Microsoft brought the semiconductor industry's own measure of success back to it on Wednesday, and argued that AI will be judged by it.

Sandisk says its NAND-based High Bandwidth Flash (HBF) technology can match HBM bandwidth while providing eight to 16 times the capacity at a similar cost, positioning flash memory closer to processors as inference workloads strain conventional memory systems.

Micron Technology is exploring tightly coupled DRAM that could deliver more than 10 times the bandwidth of HBM while using substantially less energy per bit, as the memory maker looks to new architectures to overcome the widening gap between processor performance and memory bandwidth.

At a shared stage at Semicon Taiwan 2026, executives from Samsung Electronics, SK Hynix, and Micron agreed that traditional 2.5D high-bandwidth memory (HBM) using interposers is approaching a performance ceiling. Consequently, the industry is transitioning toward 3D vertical stacking architectures to minimize transmission power consumption and leverage wafer-level bonding technologies.

Samsung Electronics and SK Hynix ended the first half of 2026 holding sharply more inventory than six months earlier, an unusual development in a memory market where demand continues to strain available supply.

Sonix Technology expects consumer IC and MCU sales to keep growing in the second half of 2026, but memory shortages could slow imaging-product shipments. AI-driven demand is also squeezing wafer foundry and packaging and testing capacity, leaving the MCU market undersupplied through the end of 2026.

Semicon Taiwan 2026 will draw more than 1,300 companies and 4,300 booths to Taipei from September 2 to 4, setting new records as the industry show expands beyond chipmaking into packaging, smart fab, quantum technology, and system integration. SEMI said the event reflects how artificial intelligence is reshaping the global semiconductor supply chain.

Apple has unveiled new Mac mini and Mac Studio models, equipped respectively with its first 2-nanometer chip, the M6, and the four-die M5 Ultra. Most notably, the latest 2-nanometer process is used only in the entry-level Mac mini, while the 3-nanometer M5 Ultra is officially billed as the "most powerful M-series chip yet."

ChangXin Memory Technologies (CXMT) has escalated its defense against the US government, shifting from supply-chain and technological competition into a full-scale courtroom battle. On August 28, 2026, the Chinese memory leader disclosed a stellar first-half financial report following its IPO and announced on the same day that it was suing the US Department of Defense (DoD).

Seven listed Chinese memory-module and memory-distribution companies have filed first-half 2026 results, and the headline numbers are the largest the tier has ever reported. The more useful number is further down each statement: four of the seven reported negative operating cash flow, as inventory purchases and other working-capital demands absorbed cash even as profits rose.

The competition in high-bandwidth memory (HBM) is moving beyond DRAM stack height, bandwidth, and packaging to the base die at the very bottom of the package. As Nvidia pushes its NVHBM architecture, the fight is shifting toward who gets to define the base die — now emerging as the control center of next-generation HBM.

Samsung Electronics is targeting mass production of its LPDDR5X processing-in-memory (PIM) solution in late 2026 or early 2027, moving the technology closer to commercialization as the memory giant broadens its portfolio beyond HBM.

Micron's plant at the Southern Taiwan Science Park, which it purchased from AUO for NT$7.4 billion (US$233.8 million) in August 2024, is now officially in operation, marking a key milestone in the transformation of its Tainan site. The move comes as Micron posts record revenue and profit, while a labor dispute at its Taiwan operations has raised strike concerns.

Nine listed Chinese memory and memory-adjacent chip companies have now filed their first-half 2026 results, and together they measure the depth of the current DRAM and NAND upcycle more precisely than any forecast. They also show how differently each company reached its number: some rode pricing, one rode an acquisition, another rode a stake revaluation, and the largest of them simply arrived at a scale nobody in the sector previously operated at.

China's leading DRAM maker CXMT has gone from years of accumulated losses to one of the most profitable companies on the mainland market, propelled by an extraordinary memory upcycle. The more consequential question is what it does with the windfall.

South Korean chip startup XCENA is pushing CXL beyond memory expansion and toward a new layer of server computing, unveiling measured results for its MX1 computational memory device at Hot Chips 2026 while expanding its work with Samsung Electronics on a rack-scale system.

Samsung Electronics has reportedly locked up as much as 70% of its memory production capacity under long-term supply agreements (LTAs), as relentless AI infrastructure demand for high bandwidth memory (HBM) tightens the broader chip shortage. In some cases, spot prices for memory products are now as much as 5 times the LTA contract price.

China's leading DRAM maker CXMT has reportedly begun small-volume production of HBM3E, a step that could ease one of the most important hardware constraints on China's domestic AI chip industry. US export controls have restricted Chinese access to advanced high-bandwidth memory (HBM), making local supply increasingly critical to scaling homegrown AI processors.

India has unveiled the next phase of its semiconductor policy, a move that could reshape global chip supply chains, investment flows, and talent markets. The expanded program aims to deepen local design and manufacturing capacity, while giving multinational companies and developers a larger role in one of technology's most strategic sectors.

Semicon Taiwan 2026 is set to open soon as Semiconductor Equipment and Materials International said AI is pushing a new wave of technology upgrades across the global chip industry. At an August 31 pre-event briefing, the group said the changes are accelerating work in AI computing, advanced process technology, heterogeneous integration, and quantum computing.

As trillion-parameter large language models (LLMs) and agentic AI become mainstream, the AI infrastructure bottleneck is shifting from raw compute power toward system-level integration across compute, memory, interconnects, and thermal management. To bypass traditional hardware constraints, Nvidia announced NVHBM, its next-generation custom HBM technology, and integrated it into the NVLink Fusion ecosystem.

Samsung Electronics is using a cross-division "One Team" model to develop its sixth-generation high-bandwidth memory, HBM4, but the approach is creating a new challenge for the newly established Device Solutions (DS) special management performance bonus system. Labor and management are now negotiating detailed payout rules to clarify how One Team participants and some contract workers will be rewarded and to prevent employees from being disadvantaged because they belong to different units.

Samsung Electronics is reportedly developing an 8-layer version of seventh-generation high-bandwidth memory for Nvidia, matching the chipmaker's requirements for its customized NVHBM platform. According to Seoul Economic Daily, the shift could give Samsung an early position in the next-generation artificial intelligence memory market if development moves ahead smoothly.

SK Hynix is reportedly considering adding Intel Foundry as a second manufacturing source for high-bandwidth memory (HBM) base dies, a move that could reduce its reliance on TSMC as the logic layer beneath HBM becomes increasingly important to cost, performance and accelerator design.