炒股就看金麒麟分析师研报，权威，专业，及时，全面，助您挖掘潜力主题机会！ 来源：Barrons巴伦 最新一期'赵何娟Talk'里，李飞飞教授认为，从"语言生成"到"世界生成"，空间智能将在两年内迎来应 用级爆发——但AI永远只是工具，选择权应该始终在我们人类手里。 作者|赵何娟 一切进展都已经比一年前大家的预期要快了很多。李飞飞也在钛媒体这期'赵何娟Talk'里对话里透露， 从"语言生成"到"世界生成"，空间智能将在两年内迎来应用级爆发。 随着2025年渐入尾声，有着"AI教母"之称的斯坦福大学教授李飞飞，带着她创立的World labs迎来了一 波又一波的新进展，包括首款商用"世界模型"Marble的发布，这开始让大家终于意识到，原来"世界模 型"并非只是概念，而已经是真实有用的。 回想我第一次见李飞飞教授，已经可以追溯到2017年，在斯坦福大学教学楼内。那一年，刚刚定居硅谷 的陈天桥先生向我和其他几位老朋友介绍了李飞飞教授，他当时特别提到：这是美国最杰出的华人科学 家之一。那时，李飞飞教授发起的ImageNet行动还在如火如荼的进行。我也第一次在与飞飞教授的见面 和交流中学到了一个新的概念：为什么是数据集 ...

朋友们，我是帮主郑重。今天，香港资本市场可能迎来一个历史性的代码：全球大模型第一股！ 主 角，是来自中国AI"六小虎"的智谱AI。 这不是普通的IPO。它背后站着清华顶尖团队，估值高达400亿。更关键的是，它过去三年营收年均增 速超过130%，毛利率保持在50%以上，展现出了难得的商业化锐度。 但这条赛道上，可不止它一位选手。几乎同时冲线的，还有风格迥异的MiniMax。如果说智谱是稳扎稳 打的"重装甲兵团"，靠服务企业和开发者扎根；那MiniMax就是灵活的"游骑兵"，其核心产品在海外青 少年中火爆，70%的收入来自海外。这是一场"学院派"与"产品派"的终极竞速。 帮主看来，无论谁先撞线，意义都已超越公司本身。这意味着资本市场首次为AGI（通用人工智能）基 座模型定价，标志着行业从"烧钱研发"进入"资本验证"的新阶段。对于我们投资者，这提供了一个观察 中国AI顶尖战力的绝佳窗口。 那么问题来了：你看好深耕底层、服务B端的"技术实力派"，还是更青睐打通海外C端市场的"产品明 星"？谁的故事更能赢得市场的长期信任？评论区聊聊你的判断。 我是帮主郑重，关注我，聚焦中长线价值。 ...

作者｜冬梅 在通用大模型赛道尚未走出"长期投入期"的背景下，资本市场已经率先迎来新的叙事节点。 12 月 21 日，通用人工智能（AGI）公司 MiniMax（稀宇科技）通过港交所上市聆讯，并首次刊发聆讯后资料集（PHIP）版本 招股书。继智谱 AI 于 12 月 19 日披露招股书后，MiniMax 成为"大模型六小虎"中第二家通过港交所聆讯的中国大模型企业， 若按计划于明年初完成上市，这家成立于 2022 年初的公司，将成为全球范围内从创立到登陆资本市场用时最短的 AI 企业之 一。 385 人历时 4 年冲刺 AGI 全球第一股，MiniMax 赴港 IPO 聆讯 MiniMax 成立于 2022 年初，被认为是国内首批大规模应用"混合专家"（MoE）架构的公司之一——这种方法后来被 DeepSeek 采用并推广。 MiniMax 在招股书中将自身定位为"生而全球化"的 AI 企业，核心目标是研发具备国际竞争力的通用人工智能模型，并面向全球 市场展开业务布局。截至 2025 年 9 月 30 日，MiniMax 持有的现金及现金等价物余额约为 10.46 亿美元，为其持续投入模型 研发和算力资源提供了较 ...

Core Insights - The launch of GPT-5 by OpenAI in August 2025 was anticipated as a significant step towards AGI but resulted in a series of crises, revealing the company's innovation stagnation and strategic shortsightedness [1][2] Group 1: Product Performance - Early users found GPT-5 did not demonstrate a qualitative leap compared to its predecessor, with some describing the updates as "mostly boring" [2] - Many users of GPT-4o protested due to a degraded experience, leading OpenAI to urgently reinstate model selection for paying users [2] - The failure to meet user expectations for a "disruptive upgrade" resulted in disappointment and backlash against the product [2][4] Group 2: Strategic Misalignment - OpenAI's attempt to reposition GPT-5.2 as a "high-efficiency executor" aimed at white-collar job needs appeared to be a reactive measure rather than a natural evolution of technology [3] - The model's capabilities remained limited to "assisting chat," failing to deliver breakthroughs in critical applications like legal document drafting and financial analysis [5] - The contradiction between elevated positioning and lagging capabilities undermined the product's credibility, as enterprise clients found GPT-5.2 unable to replace basic human roles [5] Group 3: Internal and External Challenges - The failure of GPT-5 highlighted strategic confusion within OpenAI, especially in light of competition from Google’s Gemini 3 and others, prompting a "red alert" status to focus on improving ChatGPT's user experience [7] - OpenAI's management faced turmoil during a critical IPO preparation phase, with reports of daily improvement meetings and project delays, raising doubts about its commercialization capabilities [7] - The overall journey of GPT-5 was marked by a high-profile launch followed by a chaotic conclusion, illustrating a disconnect between technological ambition and market reality [7]

在商业模式上，MiniMax将赌注押在了C端，即直接面向消费者的AI原生应用，收入占比已超70%。招股书显示，MiniMax在 2023年营收为346万美元，2024年营收为3052万美元，2025年前九个月营收为5344万美元，同比增长超过170%， 截至2025年9月30日，MiniMax已有超过200个国家及地区的逾2.12亿名个人用户以及超过100个国家的13万企业客户，其中海外 市场收入贡献占比超70%。 随着全球AGI赛道从"百模大战"进入整合期，MiniMax的上市进程备受市场瞩目。它让我们得以一窥，在ChatGPT点燃全球战火 后，中国原生AI大模型公司如何烧钱研发、攻城略地，并探索可持续的商业化道路。AI原生产品贡献增量 作为一家处于高速扩张期的研发驱动型AI公司，MiniMax的财务也呈现出高增长与高亏损并行的"典型AI叙事"。 MiniMax成立于2022年初，但从2023年才开始有营收。招股书显示，MiniMax在2023年营收为346万美元，2024年营收为3052万 美元；2025年前9个月营收为5344万美元，较上年同期的1945万美元增长175%。 其中，"AI原生产品"贡献了 ...

Group 1 - The core argument of the article is that while current AI models are becoming more powerful, the realization of Artificial General Intelligence (AGI) remains distant due to physical and resource limitations [3][22][24] - Tim Dettmers' blog post titled "Why AGI Will Not Happen" argues that due to physical constraints, meaningful superintelligence cannot be achieved [3][6][22] - The article discusses the limitations of hardware improvements and the challenges in achieving efficient computation, emphasizing that the current AI architectures are bound by physical realities [8][10][11] Group 2 - The blog highlights that the efficiency of current AI systems is far from optimal, with significant room for improvement in both training and inference processes [35][37][56] - It points out that the current models are lagging indicators of hardware development, suggesting that advancements in hardware will lead to better model performance [43][57] - The article proposes multiple pathways for enhancing AI capabilities, including better model-hardware co-design and exploring new hardware features [40][46][55] Group 3 - The article contrasts the AI development philosophies of the US and China, noting that the US focuses on achieving superintelligence while China emphasizes practical applications and productivity improvements [20][21] - It suggests that the pursuit of superintelligence may lead to difficulties, as organizations focusing solely on this goal may be outpaced by those driving practical AI applications [26][28] - The discussion includes the potential for smaller players in the AI space to innovate beyond scale, leveraging efficiency and practical applications [17][18][19]

Core Insights - The podcast featuring Sam Altman reveals his complex views on OpenAI's future, particularly regarding the necessity of going public and the implications of massive capital investments in AI infrastructure [2][9][28]. Group 1: OpenAI's Financial Strategy - OpenAI plans to invest $1.4 trillion in computing power and infrastructure over the coming years, which has raised concerns in the market [19][21]. - Altman emphasizes that this investment is not a short-term gamble but a long-term strategy validated by demand [21][27]. - The company may face losses exceeding $100 billion in the coming years, but this is a strategic choice to continue investing in training and computing power rather than a failure of the business model [28][30]. Group 2: AI Development and Competition - Altman expresses a unique perspective on the concept of AGI (Artificial General Intelligence), stating that the definition is not clearly established and that current models still have limitations [12][14]. - OpenAI operates in a "wartime" mode when competitors like Google and DeepSeek make significant advancements, indicating a constant state of urgency to improve [15][17]. - The competition is not merely about model parameters but about creating a stable platform that users can rely on, which is crucial for maintaining a competitive edge [17][19]. Group 3: The Role of Computing Power - Altman asserts that computing power is not merely a cost but a lifeline for AI development, with demand for AI capabilities expected to grow exponentially [22][24]. - The company is preparing for future demands that have yet to be fully realized, indicating a proactive approach to infrastructure development [27][33]. - Altman believes that the real risk lies in insufficient computing power, which could limit potential advancements and revenue growth [31][33]. Group 4: Market Perception and Future Outlook - Altman acknowledges the skepticism surrounding OpenAI's debt levels, but he argues that the value of AI infrastructure is unquestionable, with uncertainty mainly revolving around usage and timing [33][34]. - The company is not betting on avoiding losses but rather on the acceleration of demand for intelligence outpacing conservative expectations [33][36]. - Altman concludes that AI is a transformative technology that, once established, cannot be reversed, positioning OpenAI to lead in this irreversible trend [34][36].

本月初，OpenAI首席执行官萨姆·奥特曼(Sam Altman)在内部发出"红色警报(code red)"后，要求公司暂 停包括Sora视频生成项目在内的多个侧线项目，为期约八周，全力投入ChatGPT的改进工作，以应对日 益激烈的竞争。这一决定凸显了OpenAI内部在"追求广泛的消费者吸引力"与"实现突破性研究目标"之间 更深层次的理念分歧。奥特曼认为，为了确保OpenAI的生存，公司必须将用户满意度置于其最初追求 通用人工智能(AGI)的目标之上。 智通财经APP获悉，据媒体报道，OpenAI在人工智能(AI)运营方面效率持续提升，其算力利润率于10月 达到70%，较去年底的52%显著增长，是2024年1月的约35%的两倍。这是一个内部指标，衡量在扣除为 企业和消费级产品的付费用户运行模型的成本后营收所占的份额。这表明，尽管算力成本居高不下导致 仍未能实现盈利，但OpenAI在为支撑ChatGPT订阅业务、向企业客户出售模型使用权而投入的每一分服 务器运营成本中，正实现更高的营收回报。 据一位知情人士透露，OpenAI 算力利润率的提升得益于两大因素：一是全年算力租赁成本的持续下 降，二是公司对人工智能 ...

Core Viewpoint - MiniMax, a leading AI model unicorn, has officially passed the Hong Kong Stock Exchange hearing, signaling its IPO ambitions amidst discussions of a bubble in the AI sector, particularly concerning OpenAI and other major players [1][3]. Group 1: Company Overview - MiniMax has raised over $1.5 billion in funding within four years, attracting investments from notable firms such as MiHoYo, Alibaba, Tencent, and others [3][26]. - The company aims to develop a global general artificial intelligence (AGI) and has a presence in over 200 countries, with 70% of its revenue coming from international markets [3][29]. - MiniMax's total expenditure has been approximately $500 million, significantly less than competitors like OpenAI, which has spent between $40 billion to $55 billion [29][30]. Group 2: Technological Advancements - MiniMax is one of the few companies that have focused on multimodal model development since its inception [4]. - The company has released several models, including the M1 and M2 text models, achieving top rankings in performance metrics [8][9]. - MiniMax's speech model, Speech 01, and its subsequent versions have been recognized for their comprehensive capabilities, supporting over 40 languages and generating over 2.2 billion hours of speech [11][12]. - The video model Hailuo has also achieved high rankings in independent tests, demonstrating its advanced capabilities in video generation [13]. Group 3: Business Model and Financial Performance - MiniMax employs a "model-as-product" strategy, focusing on subscription services and cloud API sales, which contribute to a sustainable revenue model [15][18]. - The company reported revenues of $3.46 million in 2023, projected to soar to $30.52 million in 2024, marking a year-on-year increase of 782.2% [17]. - By the first nine months of 2025, MiniMax's revenue reached $53.44 million, reflecting a 175% increase compared to the previous year [17]. - The gross margin improved from -24.7% in 2023 to 23.3% in the first nine months of 2025, indicating enhanced profitability [19][20]. Group 4: Research and Development Efficiency - MiniMax's R&D expenses have increased significantly, with cloud computing costs for training rising from $4.15 million in 2022 to $142.4 million in the first nine months of 2025 [22][23]. - Despite ongoing losses, the adjusted net loss has been decreasing relative to revenue growth, suggesting a successful business model [24]. - The company has a cash reserve of $1.102 billion, sufficient to sustain operations for over 53 months without additional funding [24]. Group 5: Team Composition and Culture - Founded in 2022, MiniMax has a young team with an average age of 29, and 73.8% of its 385 employees are in R&D roles [25][28]. - The leadership team includes experienced individuals from SenseTime, contributing to the company's strong technical foundation and operational efficiency [26][27]. - The organizational structure is highly flat, promoting rapid decision-making and efficiency, with over 80% of the code being generated by AI [28][29].

MiniMax成立于2022年初，是一家"生而全球化"的AI公司，致力于研发具备国际竞争力的通用模型。截 至2025年9月30日，MiniMax持有的现金结余10.46亿美元。 12月21日消息，通用人工智能(AGI)公司MiniMax(稀宇科技)通过上市聆讯，并首次刊发其聆讯后资料集 (PHIP)版本的招股书。这是大模型"六小虎"中，继智谱AI在12月19日刊发招股书之后，第二家通过港交 所聆讯的中国大模型企业。 招股书显示，2025年前9个月营收为5343.7万美元(约3.76亿元)，较上年同期的1945万美元增长175%。 2023年营收为346万美元，2024年营收为3052万美元。 MiniMax自成立到2025年9月累计花费5亿美元(约35亿元)，对比OpenAI的400亿至550亿美元累计花销， MiniMax用不到1%的钱做了全模态全球领先的公司。 截至2025年9月底，MiniMax员工385人，平均年龄29岁，研发人员占比近74%，董事会成员平均年龄32 岁。 招股书提到，随着扩大运营规模，MiniMax的毛利率由2023年的-24.7%升至2024年的12.2%，并进一步 升至截至202 ...