News linked to both this project and an event.
Odaily News: On September 15, the South Korean semiconductor sector saw active trading. According to Gate platform data, SK Hynix (000660) is currently priced at $1,275.12, up 1.17% in 24 hours; Samsung Electronics (005930) is currently priced at $186.03, down 0.60% in 24 hours. According to CoinGlass data, Gate's related stock derivatives are actively trading: SKHX contract open interest is nearly $123 million, ranking second across the network, with 24-hour trading volume of nearly $177 million, also ranking among the top in the industry; SAMSUNG contract open interest reached $7.44 million, with 24-hour trading volume reaching $32.37 million, both ranking second across the network.Gate continues to expand its stock contract coverage, providing users with richer trading options through the industry's fastest response to hot trends and the most comprehensive target layout. Currently, Gate stock contracts already support popular stock targets from multiple markets including US stocks, Hong Kong stocks, and Korean stocks, and have further launched an innovative sector, with the first batch of 40 popular stock contracts listed, covering market hot targets such as ChangXin Memory Technologies, Unitree Robotics, and CATL, providing users with 7×24-hour diversified trading opportunities.
According to an official post from a16z crypto, a16z crypto has officially released Lattice Jolt, a new version of its open-source zkVM (zero-knowledge virtual machine) that fully transitions its underlying cryptography from elliptic curves to a lattice architecture, achieving post-quantum security while significantly boosting performance. Key highlights include: • Performance improvement: Proof generation speed increased by 2-3 times, reaching up to 2 million RV64IMAC cycles per second in CPU mode, and breaking through 10 million cycles per second with MacBook GPU acceleration (Apple Metal); • Smallest proof size: Proof footprint is under 100 KB, significantly outperforming other post-quantum zkVMs with sizes ranging from 200 KB to 600 KB; • Memory optimization: Memory consumption per cycle reduced from approximately 300 bytes to 200 bytes, supporting proof generation for millions of RISC-V cycles on mobile devices; • Security: Based on the standard Module-SIS assumption, providing a complete 128-bit security strength, sharing the same assumption framework as ML-DSA and ML-KEM; • New commitment scheme: The Akita polynomial commitment scheme, jointly developed by LayerZero in collaboration with institutions such as Carnegie Mellon University and the University of Southern California, replaces the original Dory scheme. a16z crypto also pointed out that hash-based SNARKs are widespread...
Odaily News, Vitalik posted on the X platform, expressing his hope to include the recursive STARK memory pool proposal EIP-8288 in the Ethereum I-star upgrade, stating that it can be seen as the next step after Frames. This solution can place quantum-safe signature data off-chain, reducing signature costs; the fuel cost of quantum-safe privacy transactions is also expected to drop from approximately 10 million to tens of thousands. The solution aggregates transaction dependencies through frames and recursive STARKs, with nodes sending one STARK of about 100 to 300KB per time cycle; the on-chain overhead is one STARK and 96 bytes per statement to be proven.
OpenAI officially released the GPT-6 Astra model and claimed it is nearing AGI. Overnight, markets reassessed the probability of a Federal Reserve rate hike in September at 50%, spot gold briefly surged past $4,500, and the IPO review status of Yangtze Memory Technologies Corp. changed to awaiting regulatory inquiry.
Odaily News - Coinbase has disclosed the architecture and application of its internal AI system "CEEcil." This system is not a simple chatbot but is designed as an AI agent that functions like a "team member." It can be mentioned in Slack, and possesses long-term memory, proactive follow-up, and judgment capabilities.CEEcil retains team context through a three-tier memory mechanism: in the background, it continuously extracts information such as decisions, blocked items, and owners from authorized channels; each night, it consolidates short-term observations into long-term summaries; and when answering questions, it retrieves historical memories, recent conversations, and real-time data on demand. The system also adopts a tiered model architecture, where simple queries are prioritized for processing via APIs or knowledge bases, and only complex, multi-step tasks are delegated to more powerful AI models, thereby reducing cost and latency.In practical applications, CEEcil can proactively join relevant Slack discussions, send emoji reactions, and automatically follow up when issues go unaddressed. Coinbase states that the system also once refused to submit operational documents containing customer identity information to Git, suggesting the use of redacted versions or compliant storage methods instead—demonstrating a certain level of security judgment.Architecturally, CEEcil consists of a Go service and an AI agent runtime, and connects to tools such as memory retrieval, knowledge bases, real-time queries, and Slack via MCP. Coinbase also has set up a real-time "kill switch," call limits and spending caps, audit mechanisms, and requires human review and merging of the code it generates.Coinbase says CEEcil is now able to help teams handle production incidents and answer expert knowledge that previously would have required waiting hours to obtain. In the future, the company hopes to expand this model into multiple AI agents tailored to specific teams, enabling AI not only to "answer questions" but also to continuously remember, proactively participate, and complete actual work.
According to Yonhap News Agency, SK Hynix showcased a complete lineup of storage solutions for AI infrastructure at the Dell Technologies Forum 2026 held in Seoul, including sixth-generation High Bandwidth Memory (HBM4), HBM4E, AI server DRAM, enterprise SSDs (eSSDs) and client SSDs (cSSDs). The AI data center products include liquid-cooling-ready PEB210 E1.S, high-capacity low-power PS1110 E3.S, and the company's first QLC flash-based enterprise SSD PS1101 E3.S. SK Hynix also served as a partner keynote speaker at the forum for the first time, introducing the applications of next-generation storage technologies in the AI era.
Qianhe Yibang, a 3D computing-in-memory chip company, announced that its self-developed world's first 4-layer 3D DRAM-stacked computing-in-memory chip has been successfully fabricated and powered on, marking the technology's transition from validation to engineering implementation. The chip employs a three-dimensional integrated native computing architecture, delivering a data memory access bandwidth that is an order of magnitude greater than conventional approaches. Memory access power consumption, latency, and per-processing costs are all reduced by an order of magnitude. Founded in January 2024 and incubated by NetEase, Qianhe Yibang recently closed a Series B funding round exceeding 2 billion yuan.
Odaily News: According to the official announcement, Bitget's stock perpetual contracts have been listed for CXMT (ChangXin Memory Technologies). These contracts are settled in USDT and support up to 20x leverage with 7×24 trading. As of now, Bitget supports a total of 273 stock contract instruments.
According to the official announcement, the Binance Futures platform will sequentially launch 6 USDT perpetual contracts based on traditional finance underlying assets from August 17 to 18, 2026, as follows: • GDXUSDT (Underlying: VanEck Gold Miners ETF) — August 17, 13:30 UTC • NETUSDT (Underlying: Cloudflare Class A Common Stock) — August 17, 13:35 UTC • VSTUSDT (Underlying: Vistra Corp Common Stock) — August 17, 13:40 UTC • SHOPUSDT (Underlying: Shopify Class A Shares) — August 17, 13:45 UTC • LYTEUSDT (Underlying: Roundhill Photonics & Optical ETF) — August 17, 13:50 UTC • CXMTUSDT (Underlying: ChangXin Memory Technologies, SSE 688825) — August 18, 05:00 UTC The above contracts all support up to 20x leverage, 24/7 trading, and multi-asset mode, with funding rates settled every 8 hours, capped at ±2%.
According to Korean media NATE, driven by the AI investment boom, SK Hynix's stock price has recently experienced severe volatility. Against the backdrop of intensifying market volatility, SK Group released an advertisement quoting the famous words of founder Choi Jong-geon (최종건): "Despair and hope are two sides of the same coin; despair can be turned into hope like flipping a hand," and rewrote it as "Anxiety and expectation in the AI era are also two sides of the same coin; anxiety can be turned into expectation," thereby conveying confidence in the long-term development of the AI industry. Securities firms believe that short-term stock price volatility has not changed SK Hynix's fundamentals, and the market should focus on its HBM4 technology leadership advantage and the performance stability brought by Long-Term Agreements (LTA). Hyundai Motor Securities analyst Noh Geun-chang (노근창) stated that SK Hynix's DRAM and NAND bit growth rates for the third quarter are expected to reach 9.7% and 1.5% respectively; with the expansion of HBM4 sales contribution, even if the proportion of Long-Term Agreements increases, DRAM Average Selling Price (ASP) is still expected to rise 19.9% quarter-over-quarter. Regarding competition concerns brought by China's ChangXin Memory Technologies (CXMT), Noh Geun-chang believes that considering the US continues to strengthen semiconductor equipment export restrictions and Micron is expanding domestic investment in the US, the possibility of major companies like Apple adopting Chinese memory chips is relatively low.
According to TechFlow Research, Morgan Stanley's latest research report in August indicates that the Q3 DRAM contract price quarter-over-quarter increase has fallen from an expected 20% to 15%, memory module inventory has risen to 12.5 weeks, and signs of weakness have emerged in China's consumer electronics demand. However, Morgan Stanley believes this is merely a normal cyclical slowdown, not a trend reversal. The capex growth rate of the four major cloud vendors was adjusted up from 14% to 29%, and AI demand remains strong. Samsung aims to place 60% to 70% of its capacity under long-term agreements, SK Hynix has completed LTA negotiations with approximately 10 clients, and Micron's 16 agreements cover approximately $100 billion in minimum revenue. Morgan Stanley maintains an overweight rating on Samsung and SK Hynix, with target prices of 381,000 won and 2.6 million won respectively, implying 74% to 84% upside potential. Samsung's current stock price corresponds to an expected P/E ratio of about 5x for 2026, while SK Hynix is about 5.8x. Morgan Stanley believes LTAs are changing industry earnings visibility; the market has not yet assigned a premium to LTA-supported earnings. If LTAs verify downside protection, the memory valuation system may be reshaped.
To address HBM memory supply shortages, NVIDIA is considering reducing the specifications of the next-generation Rubin Ultra GPU. According to sources cited by The Information, NVIDIA has internally tested at least three versions of the Rubin Ultra GPU with lower HBM configurations over the past few weeks, requiring HBM capacity lower than the initially announced specifications. Possible downgrade paths include scaling down from HBM4E to HBM4, and from 12Hi HBM to 8Hi HBM. This move aims to advance product launch plans under supply constraints.
Odaily News: SK Hynix and SanDisk have unveiled the first standard specification for High Bandwidth Flash (HBF), a next-generation storage technology based on NAND flash. SK Hynix presented the specification at FMS 2026, held from August 4 to 6 in Santa Clara, California, USA.HBF stacks NAND flash vertically in a manner similar to High Bandwidth Memory (HBM) to enhance capacity and data transfer speeds. The specification defines two stacked configurations—8-layer and 16-layer NAND dies—with a maximum capacity of 512GB and bandwidth offered in three tiers, supporting 0.4TB to 3.0TB per second.HBF connects to processors using the industry-standard UCIe interface, enabling compatibility with different processor types such as GPUs and CPUs. The specification has been released through the Open Compute Project (OCP), and the HBF Consortium currently includes Google and AI semiconductor company Tenstorrent.SK Hynix will also showcase its in-development 10th-generation (V10) 375-layer 4D NAND wafers and products for the first time at this event, with plans to begin mass production of high-performance, high-capacity enterprise SSDs (eSSD) utilizing this technology early next year.
According to Reuters, sources familiar with the matter revealed that CXMT is considering building a second 12-inch DRAM fab in Beijing Yizhuang and is discussing financing support with the Beijing Economic-Technological Development Area and several state-owned technology enterprises. Negotiations are currently still in the early stages, and the financing scale and structure have not yet been finalized. The proposed project may be located at the company's existing Beijing facility, but production capacity plans and total investment amount have not yet been disclosed.
Odaily News, "White-Haired Stock God" Serenity stated that Lumentum CEO Michael Hurlston has issued a warning that the indium phosphide (InP) laser supply chain for AI data centers is facing a more severe supply gap than memory chips. Even with Lumentum's five InP wafer fabs, product shipments could still fall more than 30% short of customer demand, indicating that AI infrastructure expansion is encountering a critical bottleneck in key optical communication components.Serenity pointed out that current supply pressure is mainly concentrated in the EML (electro-absorption modulated laser) segment, but the InP laser shortage could continue to spread. He remains bullish on the "bottleneck investment" opportunities in the optical communication laser supply chain, keeping an eye on companies including Applied Optoelectronics (AAOI), Sivers Semiconductors (SIVE), Lumentum (LITE), and Coherent (COHR). As AI data center capital expenditures continue to rise, optical modules, lasers, and other infrastructure segments that were previously undervalued by the market are now undergoing repricing, and the InP laser supply bottleneck could become a key limiting factor in the expansion of AI computing power.
Alibaba Cloud PolarDB and MemTensor jointly launched a one-stop AI memory solution, featuring persistent and high-availability AI memory capabilities. The solution integrates relational retrieval, vector retrieval (PGVector), and graph retrieval (PolarAGE) capabilities within a single PolarDB-PG instance, reducing P99 latency by up to 89.2%. MemTensor positions it as the MemOS persistent memory layer, emphasizing zero downtime. This collaboration directly combines database infrastructure with AI memory management, targeting AI application scenarios requiring long context or persistent state management.
citing sources familiar with the matter, that a company backed by Chinese state capital has commenced mass production of domestically developed DUV (Deep Ultraviolet) lithography manufacturing equipment. This marks a significant step forward for China's semiconductor industry in advancing import substitution.According to reports, the company plans to produce approximately 5 DUV lithography machines in 2026 and expand output to around 20 units in 2027. While there remains a gap compared to the production capacity of Dutch lithography giant ASML (which delivered 131 immersion DUV lithography systems last year), the breakthrough in mass-producing domestic DUV equipment signals further progress towards self-sufficiency and independent control of China's chip manufacturing supply chain.Sources stated that this batch of domestically produced DUV lithography machines is scheduled for delivery to major Chinese chip manufacturers, including Semiconductor Manufacturing International Corporation (SMIC), Hua Hong Semiconductor, and memory chip producer ChangXin Memory Technologies.Among these, ChangXin Memory Technologies, as a representative enterprise in China's DRAM industry, is expected to become a key adopter of advanced domestically produced semiconductor manufacturing equipment. If the domestic DUV equipment can achieve stable operation on the production lines of manufacturers like ChangXin, it will further reduce China's memory chip industry's reliance on critical overseas equipment.
Odaily reports, according to official sources, Bybit has listed the CXMTUSDT perpetual contract, supporting up to 10x leverage.CXMT (ChangXin Memory Technologies) is a leading enterprise in the DRAM memory chip field in mainland China and a core target in the A-share storage sector.
Odaily reports: trade.xyz posted on X platform, stating that ChangXin Memory Technologies has been converted to the standard XYZ perpetual contract, supporting 10x leverage and enabling 7×24 trading throughout the year.
According to TechFlow Research, Bernstein recently released the third report in its Memory LTA series, suggesting that the market's concern that LTAs might once again become "worthless paper" may be misguided. The new generation of LTAs adopts a back-end weighted margin structure, where customer default costs increase as the contract progresses, with core protection value concentrated in the latter half of the contract, coinciding precisely with the industry's downward cycle.