来源：36氪 当车圈芯片大佬进军机器人赛道。 2025春晚，一群穿着大花袄、舞着"赛博秧歌"的机器人，走进了全民视野。也就是这一晚，真正开启了人形机器人发展的元年。越来越多人开始谈论，机 器人什么时候能真正走进我们的生活。 眼下，机器人行业正从原型演示走向规模量产的阶段，由于缺乏能满足高可靠性、高集成度、高性能的"量产级"计算平台，机器人玩家们始终未能突破规 模化量产的瓶颈。而这几点恰巧是智能汽车计算芯片引领者，黑芝麻智能擅长的方向。 黑芝麻智能认为，目前机器人产业大致相当于2018年智能汽车行业的状态，但是前者的增长速度将远超后者。在智能汽车的带动下，现在的芯片产业已经 非常成熟，与此同时，涌现出诸如大语言模型、多模态大模型、世界模型等出色的AI大模型。因此，黑芝麻智能预判，机器人行业正处于爆发前夜。 基于上述判断，黑芝麻智能入局机器人赛道，填补了这一关键空白。 与此同时，具身智能行至商业化的路口，凭借在汽车行业积累的芯片研发基础以及供应链体系化能力，黑芝麻智能无疑是具身智能商业化落地的有力推 手。可以说，黑芝麻智能的入局，恰逢其时。 机器人，困在量产前夜 1959年，美国发明家乔治・德沃尔与约瑟夫・恩格尔 ...

基于上述判断，黑芝麻智能入局机器人赛道，填补了这一关键空白。 与此同时，具身智能行至商业化的路口，凭借在汽车行业积累的芯片研发基础以及供应链体系化能力，黑芝麻智能无疑是具身智能商业化落地的有力推手。 可以说，黑芝麻智能的入局，恰逢其时。 眼下，机器人行业正从原型演示走向规模量产的阶段，由于缺乏能满足高可靠性、高集成度、高性能的"量产级"计算平台，机器人玩家们始终未能突破规模 化量产的瓶颈。而这几点恰巧是智能汽车计算芯片引领者，黑芝麻智能擅长的方向。 黑芝麻智能认为，目前机器人产业大致相当于2018年智能汽车行业的状态，但是前者的增长速度将远超后者。在智能汽车的带动下，现在的芯片产业已经非 常成熟，与此同时，涌现出诸如大语言模型、多模态大模型、世界模型等出色的AI大模型。因此，黑芝麻智能预判，机器人行业正处于爆发前夜。 机器人，困在量产前夜 1959年，美国发明家乔治・德沃尔与约瑟夫・恩格尔伯格成功研制出世界上首台工业机器人"Unimate"，并于1961年在通用汽车工厂正式投入使用。自此， 机器人行业完成了初期探索，并正式从实验室走向工业应用领域。 经过数十年的漫长发展，如今，机器人产业正迈入"多样化需求驱动 ...

Core Insights - The memory supply chain is experiencing a surge driven by strong AI demand, leading to a price increase cycle for memory chips, with mobile DRAM contract prices expected to rise by 30%-40% in Q4 2025 and NAND prices increasing by a high single-digit percentage, potentially continuing into mid-2026 [1][2] Group 1: Impact on Different Smartphone Segments - The impact of rising memory prices varies significantly across smartphone segments, with mid-range and low-end models like the Redmi series being the most affected, where memory costs account for over 10% of ASP, potentially leading to a 2-3 percentage point decline in gross margins for Xiaomi [1][3][4] - High-end models, such as the iPhone, are less affected as memory costs constitute only 4% of ASP, indicating a stronger resilience against price hikes [3][4] Group 2: Market Dynamics and Manufacturer Strategies - The surge in AI demand is fundamentally different from past price fluctuations driven solely by supply-demand cycles, with AI servers requiring eight times the DRAM and three times the NAND compared to regular servers, prompting major chip manufacturers like Samsung and SK Hynix to shift production focus from low-margin LPDDR chips to high-margin HBM products [2][6] - The ongoing supply tightness is exacerbated by chip manufacturers pausing quotes, creating a dilemma for smartphone manufacturers caught between the risks of purchasing at inflated prices or facing shortages [2][6] Group 3: Historical Context and Future Trends - Historical trends indicate that memory price increases often lead to market consolidation, with smaller brands struggling to adapt and potentially exiting the market, while leading companies gain market share [7] - The current environment is prompting manufacturers to adopt new survival strategies, including high-end product line expansions, improved supply chain management, and technological innovations to mitigate cost pressures [6][7]

Core Viewpoint - Concerns about an AI bubble are not primarily an issue for Nvidia, but rather for companies that are heavily borrowing to build data centers, which may face liquidation in two to three years when capacity becomes saturated [1][3]. Group 1: Nvidia's Performance and Market Sentiment - Nvidia's revenue and forecasts have exceeded market expectations, with CEO Jensen Huang stating that the situation observed is different from the AI bubble narrative [3][6]. - Nvidia has secured $500 billion in orders for advanced chips before 2026, indicating strong demand from major clients like Microsoft, Amazon, Google, and Meta [3][4]. - Some analysts believe that the strong performance of Nvidia only reflects robust infrastructure spending and does not indicate the true maturity of the AI economy [3][6]. Group 2: Debt and Data Center Concerns - Analysts warn that the real risk lies in the financing model of data centers, which are often funded through significant debt by major cloud service providers [5][6]. - The speculative nature of data center investments may lead to challenges when global capacity reaches saturation in two to three years [6][8]. - Concerns are raised about the thin revenue of AI developers like OpenAI compared to their substantial expenditures, which may unsettle investors [6][8]. Group 3: Market Dynamics and Future Outlook - Despite potential challenges for AI startups, Nvidia is expected to continue selling products to large cloud service providers and sovereign AI projects, supporting its market valuation [7][8]. - Analysts express a divided view on whether the current infrastructure boom is sustainable or indicative of a bubble, with some seeing Nvidia's results as a positive signal for long-term growth in AI demand [8][9]. - Nvidia's CEO has countered the AI bubble narrative, emphasizing a different perspective on the market's trajectory [9].

Group 1 - The core viewpoint is that despite warnings about a potential AI bubble, it is not a good time to short major US tech companies like Nvidia, according to Muddy Waters Capital CEO Carson Block [1] - Nvidia's recent earnings report showed a record revenue of $57 billion for the October quarter, a 62% increase year-over-year, surpassing analyst expectations [1] - Nvidia raised its revenue forecast for the current quarter to $65 billion, exceeding previous analyst estimates of $62.1 billion [1] Group 2 - Nvidia's quarterly earnings reports are viewed as a significant indicator of the health of the tech industry and the overall market, with 45% of global fund managers considering AI stock market bubbles a major risk [2] - Block is focusing on smaller companies involved in AI for potential shorting opportunities, indicating that many companies are either genuinely involved in AI or are "pretenders" [3] - The passive trading boom has disrupted the market and weakened price discovery, with funds continuously buying Nvidia as long as there is inflow, regardless of its price [3]

人民财讯11月20日电，摩尔线程(688795)11月20日公告，发行人和保荐人（主承销商）根据初步询 价结果，协商确定本次发行价格为114.28元/股，网下发行不再进行累计投标询价。发行人本次募投项目 预计使用募集资金为80亿元。按本次发行价格114.28元/股和7000万股的新股发行数量计算，预计发行人 募集资金总额为79.996亿元，扣除发行费用，预计募集资金净额为75.76亿元。 标签： 财经频道 财经资讯 (资料图片仅供参考) ...

新京报贝壳财经讯 11月20日，摩尔线程公告，首次公开发行股票并在科创板上市，发行股份数量为 7000万股，全部为公开发行新股。本次网上发行与网下发行将于2025年11月24日(T日)分别通过上交所 交易系统和互联网交易平台实施。发行价格为114.28元/股，预计募集资金总额为80亿元，扣除发行费用 后，预计募集资金净额为75.76亿元。 ...

尽管英伟达的业绩被广泛视为AI产业健康的重要指标，但多位分析师指出，真正的风险并非来自芯片 制造商本身，而是那些举债建设数据中心的企业。 据华尔街见闻文章，这家AI芯片巨头公布的营收和预期均超出市场预期，其CEO黄仁勋在财报电话会 议上明确表示"我们看到的情况截然不同"，试图打消AI泡沫论。英伟达披露其先进芯片在2026年前已获 得5000亿美元订单，微软、亚马逊、谷歌和Meta等主要客户已明确计划加速AI芯片采购。 然而，11月20日，据报道，D.A. Davidson科技研究主管Gil Luria警告称，对人工智能泡沫的担忧并非英 伟达的问题，真正的问题在于企业为了建设数据中心而举借大量债务。他指出，数据中心本质上是投机 性投资，可能在两到三年后全球产能达到饱和时面临清算。 有分析师指出，英伟达的亮眼业绩只能证明AI基础设施支出旺盛，但并不能说明AI经济已真正成熟。 更关键的指标是微软、Adobe等平台上AI服务的实际变现能力和客户需求，而非单纯的芯片销售数据。 不过，也有分析师认为，AI基础设施需求持续超过供给，并引用英伟达2026年前5000亿美元订单称"这 只是开始"。Wedbush Securit ...

新华财经上海11月20日（记者魏雨田）三季度全球资本市场波动加剧，随着美股机构投资者三季度13F 报告披露陆续收官，头部投资机构的调仓路径正式浮出水面。从调仓逻辑来看，科技与消费龙头成为机 构加仓的核心标的，而英伟达则在机构间引发分歧。当前市场围绕"AI泡沫"的争论愈演愈烈，不过机构 普遍认为，当下AI领域的高预期实为产业升级的重要催化剂，且技术商业化进程已实质性启动，具备 核心优势的头部平台将持续受益。 加仓聚焦核心标的科技消费龙头获青睐 三季度以来，全球资本市场波动加剧，多家头部投资机构将加仓重心明确指向科技及消费领域的龙头企 业。 从知名投资人段永平旗下H&H International Investment公布的13F文件来看，尽管该机构对苹果进行了小 幅减持，但这一动作并未动摇苹果的核心持仓地位。截至三季度末，苹果依旧是其第一大重仓股，持仓 占比达60.42%，对应持仓市值约88.69亿美元。在减持苹果的同时，段永平三季度对伯克希尔·哈撒韦进 行了幅度超53%的增持，目前持仓占比为17.78%，持仓市值为26.1亿美元，展现出段永平对伯克希尔·哈 撒韦投资逻辑的坚定信心。 机构持仓分歧凸显 AI龙头 ...

证券日报网讯北京君正（300223）11月20日在互动平台回答投资者提问时表示，公司智能视觉芯片主要 面向智能安防、泛视觉等领域，无需3D技术。 ...