（文章来源：每日经济新闻） 每经AI快讯，2月3日，英伟达持仓股美股盘前走强，CoreWeave涨2.6%，Nebius涨2.8%，Applied Digital 涨2.9%。 ...

Core Insights - The artificial intelligence narrative is becoming more intense due to concerns raised by OpenAI regarding the quality of Nvidia's microchips [1] Group 1: Company Concerns - OpenAI has expressed worries about the performance and reliability of Nvidia's microchips, which are critical for AI applications [1] - This concern follows a public appearance by Nvidia's CEO Jensen Huang, indicating potential issues in the supply chain or product quality [1]

Core Insights - Nvidia and OpenAI are attempting to navigate a complex relationship, with recent reports of a breach being viewed with skepticism [1] Company and Industry Summary - Nvidia is actively working to maintain a balanced partnership with OpenAI amidst challenges [1] - The relationship dynamics between Nvidia and OpenAI are under scrutiny, particularly in light of potential security concerns [1]

Core Viewpoint - Spain's competition authority has imposed fines totaling 20.5 million euros ($24.2 million) on three companies within the Repsol group for engaging in what is described as an abusive margin-squeeze strategy [1] Group 1: Regulatory Actions - The fines are a result of the companies' practices that the competition watchdog deemed harmful to market competition [1] - The total amount of fines levied is significant, indicating the authority's strict stance on anti-competitive behavior [1] Group 2: Impact on Repsol Group - The financial penalty may impact the Repsol group's operational costs and overall profitability [1] - This regulatory action could lead to increased scrutiny of the Repsol group's pricing strategies in the future [1]

Shares of Samsung Electronics and SK Hynix surged Tuesday as investors took the prior session's selloff as an opportunity to pile back in to the chip-making giants. ...

 VANCOUVER, BC - TheNewswire - January 3, 2026 – Rush Gold Corp. (the “Company” or “Rush”) (CSE: RGN; FSE: B6H; OTCQB: RGNCF) is pleased to announce that its common shares have commenced trading on the OTCQB® Venture Market in the United States under the symbol “RGNCF”. Rush’s common shares will continue to trade on the Canadian Securities Exchange under the symbol “RGN”. The Company has also secured eligibility for its common shares to settle electronically via the Depository Trust Company (“DTC”), enabli ...

2月3日消息，OpenAI 首席执行官萨姆・奥尔特曼（Sam Altman）今日在 X 平台（前身为推特）发文 称："我们喜欢与英伟达合作，他们生产世界上最好的人工智能芯片。我们希望长期成为他们的大客 户。我实在搞不懂这些毫无根据的疯狂传言是从哪里来的。" 奥尔特曼此番表态前，路透社曾报道称这家由微软投资支持的 ChatGPT 开发商，正在寻求英伟达人工 智能芯片的替代方案；另有《华尔街日报》消息称，英伟达计划向 OpenAI 投资 1000 亿美元的方案已 陷入停滞。 去年 9 月，英伟达表示计划向 OpenAI 注资高达 1000 亿美元，作为交易的一部分，这家芯片制造商将 获得这家初创公司的股份，而 OpenAI 则将获得购买先进芯片所需的资金。路透社报道，这笔交易原本 预计在几周内完成。但实际上，谈判已经拖延了数月。在此期间，OpenAI 与 AMD 达成了多项协议。 上周六，英伟达首席执行官黄仁勋对有关其与 OpenAI 关系紧张的报道予以驳斥，称这种说法是"无稽 之谈"，并表示英伟达计划对 OpenAI 进行大规模投资。 ...

最近有消息称OpenAI对英伟达的某些最新人工智能芯片表示不满，并且自去年以来就一直在寻求替代 方案，这可能会使这两家在人工智能热潮中最为知名的公司之间的关系变得有些复杂。 尽管英伟达CEO黄仁勋和OpenAI CEO阿尔特曼都隔空反驳了这一传言，不过这一传言凸显出，虽然英 伟达在用于训练大型人工智能模型的芯片方面仍占据主导地位，但AI推理可能已成为芯片竞争的新战 场。 英伟达芯片的推理短板？ 最近有消息称，由于OpenAI的策略转变，使得其越来越重视用于执行人工智能推理部分的芯片。AI 推 理是指 AI 模型（例如 ChatGPT 应用所使用的模型）响应用户查询和请求的过程。 毋庸置疑的是，英伟达的GPU非常适合用于处理大规模数据运算，这是训练像 ChatGPT 这样大型人工 智能模型所必需的。但如今，人工智能的进步越来越侧重于将训练好的模型用于推理和判断，这可能是 一个全新的、更大的的人工智能阶段。 推理所需的内存比训练更多，因为芯片需要花费相对更多的时间从内存中获取数据，而不是执行数学运 算。 然而，英伟达和AMD的GPU技术依赖于外部内存，这会增加处理时间，并减慢用户与聊天机器人互动 的速度。 相比之 ...

Core Viewpoint - Nvidia is expected to continue dominating the AI infrastructure market in 2026, driven by strong demand for its Blackwell systems and increased spending on AI data centers [1][2]. Group 1: Revenue Predictions - Nvidia is projected to surpass the consensus revenue estimates of $323.3 billion for fiscal 2027, supported by multiple growth catalysts [3]. - The company has revenue visibility exceeding $500 billion for its Blackwell and next-generation Rubin systems from early 2025 through 2026 [5]. Group 2: Market Dynamics - Hyperscalers are shifting towards rack-scale solutions that integrate GPUs, CPUs, networking, and software, enhancing Nvidia's pricing power [5]. - The transition from infrequent training workloads to repetitive inference workloads is expected to drive demand for Nvidia's newer platforms, such as the Vera Rubin systems, anticipated to launch in the second half of 2026 [6]. Group 3: Profitability and Margins - Nvidia is expected to maintain gross margins around 75% in fiscal 2027, supported by a revenue mix focused on high-margin data center GPUs and networking products [8]. - The company is also experiencing growth in its high-margin software and services business, which contributes to its profitability [8]. Group 4: Market Share and Competitive Position - Nvidia held a 92% share of the global GPU market at the end of Q3 fiscal 2025, despite a slight decline due to competition [10]. - The company's competitive advantage extends beyond hardware to its CUDA software ecosystem and developer tools, which facilitate faster deployments and increase switching costs for customers [11].

【文/观察者网 阮佳琪】 前脚刚有消息称英伟达CEO黄仁勋对OpenAI心存不满、千亿投资谈判迟迟没有进展，后脚OpenAI又被 曝出"看不上"英伟达的人工智能（AI）芯片，这一系列动向让AI热潮中最受瞩目的这两家美国企业之间 的关系愈发复杂。 当地时间2日，路透社援引多位知情人士消息称，OpenAI对英伟达最新推出的部分AI芯片在推理环节的 表现感到不满（即模型响应用户查询和请求的过程），自去年起已着手寻找替代方案。 据七名消息人士透露，OpenAI的核心不满集中在速度层面：英伟达硬件在处理软件开发、AI系统间通 信等特定问题时，向ChatGPT用户输出答案的响应速度未达预期。 报道称，OpenAI的战略转向，愈发重视用于执行AI推理特定环节的芯片，而传统英伟达GPU依赖外部 存储的架构，与推理任务需频繁读取数据的需求存在矛盾，反观Anthropic、谷歌等竞争对手，早已依 托谷歌自研的张量处理单元（TPU）占据优势。这类专为推理计算设计的芯片，性能表现优于英伟达的 通用GPU。 一名消息人士补充，这一问题在OpenAI大力推广的代码生成产品Codex上体现得尤为突出，公司内部员 工也将Codex的部分性能 ...