Chinese supercomputer maker Sugon and chip designer Hygon Information Technology have called off a planned mega-merger that had been under discussion for months, in a surprise move that could complicate the country's tech self-sufficiency efforts in the semiconductor and high-performance computing sectors. The two Shanghai-listed companies late on Tuesday said they cancelled merger talks because "the market environment has changed significantly since the initial planning stages of this transaction". They ...

Core Insights - A class action lawsuit has been filed against Bitdeer Technologies Group for allegedly misleading investors about its business prospects and the development of its SEALMINER A4 machine [1][2] Allegations - The complaint claims that Bitdeer misrepresented its progress on the SEAL04 ASIC chip technology, which was expected to enhance energy efficiency and increase revenue through self-mining and external sales in fiscal year 2025 [2] - It is alleged that the company faced issues with the SEAL04 chip design, leading to delays in production and a dual-track approach to ensure mass production [2] Financial Performance - On November 10, 2025, Bitdeer reported a net loss of $266.7 million, or $1.28 per share, primarily due to increased operating expenses related to R&D of its ASIC roadmap [3] - Following the announcement of these financial results, Bitdeer's stock price fell over 14%, from $17.65 per share to $15.02 per share [3] Class Action Participation - Shareholders interested in serving as lead plaintiffs must file their papers by February 2, 2026, but participation is not required to be eligible for recovery [4] Company Background - Robbins LLP is noted for its focus on shareholder rights litigation, aiming to help shareholders recover losses and improve corporate governance since 2002 [5]

Core Viewpoint - A class action has been filed against Bitdeer Technologies Group, alleging that the company misled investors regarding its business prospects and the development of its SEALMINER A4 machine [1][2]. Allegations - The complaint claims that Bitdeer created a false impression of having reliable information about the SEAL04 ASIC chip technology, which was expected to enhance revenue through self-mining and external sales in fiscal year 2025 [2]. - It is alleged that the company misrepresented the progress of the SEAL04 chip design, which ultimately led to delays in production and a dual-track approach to ensure mass production [2]. Financial Results - On November 10, 2025, Bitdeer announced a net loss of $266.7 million, or $1.28 per share, primarily due to increased operating expenses related to R&D of its ASIC roadmap [3]. - Following the announcement, Bitdeer's stock price fell over 14%, from $17.65 per share to $15.02 per share [3]. Class Action Participation - Shareholders may be eligible to participate in the class action against Bitdeer, with a deadline to file as lead plaintiff by February 2, 2026 [4]. - Shareholders can choose to remain absent class members without participating in the case [4]. Company Background - Robbins LLP is noted as a leader in shareholder rights litigation, focusing on helping shareholders recover losses and improve corporate governance since 2002 [5].

Core Insights - Martin Shkreli has made a strategic investment in Q/C Technologies Inc. (NASDAQ:QCLS) with a price target of $100, emphasizing the potential of photonics over quantum mechanics for high-performance computing [1][2] - Shkreli argues that photonic systems can outperform current GPUs in AI tasks, labeling the technology as "quantum class" [2] - A recent Microsoft study supports Shkreli's thesis, showcasing an Analog Optical Computer (AOC) that could achieve over 100 times the efficiency of leading GPUs [3][4] Investment Thesis - Shkreli's investment is based on a Microsoft paper that discusses the AOC's capabilities in AI inference and optimization, predicting that major companies will adopt optical computing within the next 5-10 years [3][4] - The AOC utilizes a dual-domain approach, combining analog electronics and 3D optics, to enhance efficiency by avoiding energy-intensive digital conversions [4][5] Technological Validation - The Microsoft study demonstrates the AOC's application in real-world scenarios, such as medical imaging and financial transactions, indicating its practical viability [5] - By leveraging light intensity for data encoding and calculations, the AOC addresses the limitations of current digital chips, signaling a potential hardware transition in the market [6] Company Performance - QCLS's stock price increased by 38.17% to $4.67, with a further rise of 3.43% in after-hours trading; however, it has seen a significant decline of 95.94% year-to-date [7]

1990年，为推动"863计划"项目落地，"国家智能计算机研究开发中心"成立，中科院计算所研究员李国杰出任了中心主任。 经历过"玻璃房"屈辱的李国杰，上任立的第一个项，就是造超算。 重器 2013年岁末，月球虹湾。 "嫦娥三号"轻盈降落，举世沸腾。 完美一落的背后，是世界顶级难题的攻克。 美、苏航天器曾多次着陆月球，有的因为错过窗口期，冲入深空；有的因姿态失控，上演了"撞月"的悲剧……总而言之，上去难，着陆落地更难。 它是世上最难的计算题之一。从轨道修正到姿态调整，每个环节都要毫秒级实时运算作支撑，它一旦"差之毫厘"，便真会"谬以千里"。 "嫦娥三号"能够平稳着陆，离不开支撑其运算的硬核重器——中科曙光的高性能计算机（HPC）。 你可能没听说过"曙光"，但一定"见识"过它的HPC。 从气象预报，到石油探测；从证券交易，到"神舟"飞天、"嫦娥"落月……到处都有中科曙光HPC的身影。 所谓HPC，俗称"超算"。 个人电脑一般只有一个处理器，超算却能将成千上万枚处理器汇聚一体，实现磅礴算力，处理天量数据。 从太空探索、到模拟核试验，从天气预报、到人类基因测序……没有超算，这些超级工程寸步难行。 计算驱动的科技时代， ...

Core Insights - Penguin Solutions, Inc. is gaining attention as a small-cap tech stock on Wall Street [1] - The company announced the general availability of Oracle Linux OS on its Stratus ztC Endurance platforms, achieving a system reliability of seven nines (99.99999%) [2] - This reliability translates to a maximum of 3.15 seconds of unplanned downtime or data loss per year on a single platform [2] Company Overview - Penguin Solutions specializes in high-performance computing, AI infrastructure, fault-tolerant platforms, and advanced memory solutions [4] - The Stratus ztC Endurance architecture allows for the consolidation of Oracle workloads without complex clustered server setups, enhancing predictive fault tolerance [3] Market Positioning - The solution is targeted at sectors requiring always-on environments, including energy, financial services, retail, telecommunications, and government [3] - Support for Oracle Linux OS on Stratus ztC Endurance is expected to be available in November 2025 [3]

Core Insights - Hewlett-Packard Enterprise Company (NYSE:HPE) is gaining attention on Wall Street as an AI stock, with a price target raised to $30.00 from $29.00 while maintaining a Strong Buy rating [1][3] - The company's F3Q25 earnings report showed a mixed performance, with expectations of a 30% quarter-over-quarter decline in AI platform sales, which is considered normal [3] - The inclusion of Juniper is positively impacting margins and is expected to enhance HPE's role in AI initiatives [2][3] Financial Performance - HPE's recent earnings report reflected a beat, but guidance remains mixed, indicating a need for investor patience to appreciate long-term potential [3] - The operating margin outlook is improving due to Juniper's contribution [2][3] Strategic Outlook - An analyst meeting is scheduled for October, where HPE is expected to disclose its strategy and financial outlook [2][3] - HPE is positioned within the context of an AI networking basket, indicating a broader market engagement [3]

Core Insights - The CoDesign 2025 International Symposium held in Osaka, Japan, focused on the integration of high-performance computing (HPC) and artificial intelligence (AI) [1][2] - The symposium addressed four core areas: algorithms, application systems, system software and middleware, and hardware-software co-design [1] - Key discussions included the future of exascale computing and the role of HPC and AI in advancing scientific research [1] Group 1: Key Presentations and Discussions - Professor Lu Yutong highlighted the challenges of system fragmentation affecting computational efficiency and emphasized the importance of co-design in hardware and software [1] - Shuaiwen Leon Song introduced the Together AI's "AI Accelerated Cloud" platform, showcasing its self-developed inference engine and optimization strategies [1] - Professor Thomas C. Schulthess presented the cloud-native supercomputing platform ALPS developed by CSCS, which supports elastic resource scheduling and the "science as a service" concept [2] Group 2: Research Focus Areas - Professor Xian-He Sun addressed the "memory wall" issue, proposing theories and models to optimize data flow and enhance performance [2] - Experts shared advancements in large model training optimization, supercomputer architecture, scheduling algorithms, memory problem solutions, and data compression tools [2]