TrendForce's latest research reveals that the global DRAM industry revenue reached approximately $154.73 billion in Q2 2026, rising 59.5% quarter-over-quarter, driven by AI server demand for HBM3e, LPDDR5X, and high-capacity RDIMM. Supply growth remains constrained as manufacturers prioritize server allocation, with bit shipments expanding only modestly, signaling that AI demand continues to exceed supply expansion.

A research note from Schulz_Research said China’s memory buildout is no longer centered on a single company, but on a three-part structure led by CXMT, YMTC and XMC. In the note, CXMT is described as handling DRAM wafers, while YMTC focuses on NAND and is adding DRAM at its newest plant. XMC, a foundry controlled by YMTC, is tasked with stacking products from both lines. The post linked that structure to rules Beijing has applied since late December last year. Under those rules, newly approved factories must show in procurement tenders that at least half of their equipment is sourced domestically, with exemptions granted only when no domestic option is available. Schulz_Research said YMTC’s Wuhan Phase 3 is the first advanced memory project cleared under that framework and is set to begin production later this year. The note also outlined capacity figures for CXMT, YMTC and XMC, alongside market-share estimates and product updates including DDR5, LPDDR5 and planned HBM3 production.

PA Daily’s latest roundup spans crypto markets, regulation, AI, funding, and whale activity. OpenAI formally introduced GPT-6 Astra and said the model may sit close to the point where artificial general intelligence becomes recognizable in hindsight. Nvidia, meanwhile, agreed to acquire AI platform Hugging Face for about $13 billion, a deal that would put its three French co-founders into the billionaire ranks while preserving Hugging Face’s open-platform model, according to the report. In crypto, spot Bitcoin ETFs posted $731 million in net inflows on Sept. 3 U.S. time, led by BlackRock’s IBIT at $454 million. Spot Ether ETFs added another $141 million. Polymarket launched perpetual contracts across crypto, equities, and commodities, while Ethena said its fee switch proposal passed unanimously, clearing the way for programmatic ENA buybacks. The report also tracks a sharp response from AMC CEO Adam Aron, who said Robinhood’s tokenized stock plan involving AMC and more than 190 companies was unauthorized and unregistered under U.S. securities law. Elsewhere, a Guangzhou court ruled that a virtual-currency lending contract was invalid and said losses should be calculated using acquisition cost. Binance Alpha listings, funding for Diameter Pay, Oura’s U.S. IPO filing, and a series of high-return meme-token trades also featured in the day’s developments.

ByteDance and ZTE have jointly developed the Nubia NaviX Ultra, a 'Doubao' phone featuring ChangXin's highest-speed LPDDR5X memory at 10667Mbps, marking the first mass production of domestic high-speed LPDDR5X memory chips. The phone has completed network access registration and is expected to enter mass production in September.

An analyst at Citrini said Nvidia's Rubin Ultra could switch from 12-layer HBM4E to 8-layer HBM, citing sources. OpenAI and Anthropic had asked for 4-layer parts but memory makers rejected the idea, so the cut may stop at 8 layers. Jukan, the analyst, attributes the move mainly to yield and cost concerns: with 12-layer HBM4E and price increases, memory could account for roughly 70% of Rubin Ultra's total bill of materials. Software optimizations such as model quantization, MLA, and computation splitting are moving rarely accessed KV cache and model states to LPDDR, CXL, and NAND, leaving HBM for the active working set. After minimum capacity is met, customers now value bandwidth over capacity. Fewer layers could improve packaging yield and lift HBM and AI accelerator shipments, potentially expanding total HBM demand. Higher bandwidth requirements also reduce the share of wafers passing speed binning, consuming extra DRAM wafer capacity. Longer term, HBM will be replaced by new architectures, likely merging memory and logic; the next two years are critical for memory makers to expand into logic.

WuBlockchain’s WhiteLine Daily highlighted four developments on Aug. 29. First, after Federal Reserve Chair Kevin Warsh spoke at the 2026 Jackson Hole global central banking symposium, market pricing for a September rate hike rose from about 35% the previous day to nearly 60%. Warsh said the Fed may still need to act if underlying inflation does not return to the 2% target quickly enough, while noting a 4.1% unemployment rate, financial conditions that are hard to call restrictive, and a 3.7% reading for the Fed’s preferred PCE inflation gauge. Second, Changxin Memory announced formal mass production of its next-generation LPDDR6 memory, with Xiaomi 18 Fold set to be the first device to carry it. Third, laser chip maker Yuanjie Technology reported strong first-half 2026 results, with data center revenue up 640.29% year over year to 774 million yuan, accounting for about 84% of total revenue. Fourth, OpenAI said it plans to end its direct model supply contract with AI coding tool Cursor on Nov. 12, while Reuters reported that Anthropic will increase Claude model and compute support for Cursor.

ChangXin Technology said on August 29 that its self-developed LPDDR6, the latest generation of low-power memory, has entered mass production and will first be used in Xiaomi’s 18 Fold flagship foldable phone. The company described the launch as the first commercial deployment of an LPDDR6 product globally. According to ChangXin’s half-year report, the LPDDR6 chip delivers broad performance gains over the previous LPDDR5X generation. The company said the product uses architectural changes and data-transfer optimization, reaching a peak speed of 12,800 Mbps and a maximum chip capacity of 16 GB. LPDDR6 is the newest international LPDDR standard. JEDEC first released the LPDDR6 specification in July 2025. Market chatter in March had already indicated that ChangXin’s LPDDR6 products were being sent for testing. The company said it moved from standards alignment to sample delivery, validation, mass production, and commercialization, culminating in what it called a global first launch. The report also said LPDDR6 is seen as a key building block for on-device AI computing, with use cases spanning smartphones and PCs, and extending to smart cockpits and AI data centers.

ChangXin Technology released its 2026 interim report, posting H1 revenue of RMB 150.31 billion, up 873.64% year-on-year, and net profit attributable to shareholders of RMB 77.61 billion, reversing a net loss of RMB 2.33 billion a year earlier. Operating cash flow reached RMB 131.16 billion, surging 2,985.64%. Gross margin stood at 84.84%, with R&D spending of RMB 6.86 billion, or 4.56% of revenue. Total assets reached RMB 468.08 billion, with net assets of RMB 270.75 billion. The company, the world's fourth-largest and China's largest DRAM IDM, supplies DDR5, LPDDR5/5X and LPDDR6 products to clients including Alibaba Cloud, ByteDance, Tencent, Lenovo, Xiaomi, OPPO and vivo. As of June 30, it held 4,484 domestic and 3,400 overseas patents, with 7,491 R&D staff accounting for 33.42% of headcount. ChangXin has no controlling shareholder or de facto controller; its largest shareholder, Qinghui Jidian, holds a 21.67% stake. The company listed on the STAR Market on July 27 at RMB 8.66 per share, with market cap surpassing RMB 3.5 trillion on its first trading day.