2026年新年第一天，DeepSeek又开卷了。 发了他们新年的第一篇论文。 《 mHC: Manifold-Constrained Hyper-Connections 》 感觉是DeepSeek-V4的铺垫，当然一些小道消息，不保真，我也不懂，我只是拍脑袋预测一下，有问题别找我。 就是V4，大概在1月中下旬或者1月底，然后呢，有多模态输入，没有多模态输出。 就酱，回到论文。 这篇论文我是说实话，有点过于硬核了。 但同时，传递出来的信息量和对AI界的改变，又是巨大的。 在给自己放了一天假，然后啃了一天以后（这玩意比我想象的难啃多了。。。）我还是想，用最通俗易懂最有意思的方式，来跟你聊聊，这篇论文的有 趣之处，以及，是如何对现在的生态进行一些新的输入的。 当然也给我自己叠个甲，我不是算法出身，我只是读完以后觉得很棒想分享给大家看，我对这篇论文的理解和乱七八糟的各种名词解释，都是我自己民 科瞎JB自学的，部分措辞也有为了能让大家更好理解而做的部分简化，如果有我理解的错误或者事实性错误的地方，欢迎大佬们在评论区指正讨论，感 谢。 话不多说，我们，正是开始。 在最开始之前，我想先问大家一个问题，就是大家认为，一个要处 ...

Group 1 - France's Paris prosecutor has confirmed an investigation into Elon Musk's AI company xAI regarding its chatbot "Grok" for allegedly generating illegal pornographic content, extending an existing investigation into the X platform since July last year for foreign interference assistance [2] - MSI has announced a pre-release of its gaming monitor MEG X, claiming it to be the world's first AI esports monitor equipped with next-generation QD-OLED technology, featuring a wider aspect ratio than 16:9 [2] Group 2 - DeepSeek has published a paper proposing a more efficient AI development method, introducing a framework called "manifold-constrained hyperconnection" (mHC) aimed at enhancing scalability while reducing computational and energy demands for training advanced AI systems [5] - The world's first visual AI tennis robot, Tenniix, is set to debut at CES 2026, featuring smart tracking, adaptive learning, and human-like training capabilities, with a starting price of $699 [5]

Core Insights - Sichuan Province has issued a three-year action plan to empower new industrialization with artificial intelligence, marking the first provincial-level initiative in the western region to promote AI applications in this field [1] - The plan aims to enhance technological innovation, promote smart product development, and accelerate the integration of AI into key industrial chains, targeting a revenue of over 130 billion yuan from the AI industry by 2024, with a year-on-year growth of over 25% [1] Group 1 - The action plan outlines five key tasks, including strengthening technological innovation, upgrading product iterations, driving application scenarios, establishing governance mechanisms, and building industrial ecosystems [1] - By 2027, the plan aims for full digital transformation coverage among large-scale industrial enterprises, the establishment of over 50 digital industrial parks, and the cultivation of more than 90 national champions and "little giant" enterprises in the AI sector [1] Group 2 - Sichuan will accelerate the construction of a self-innovation system for the entire AI chain, providing up to 10 million yuan in funding support for high-level software and hardware technology breakthroughs [2] - The province will address issues of shallow integration of AI with industry and fragmented application promotion by opening over 20 application scenarios annually and promoting the use of vertical large models in various industrial processes [2] - To support high-quality industrial development, Sichuan will enhance the infrastructure for computing power, networks, and data, aiming for comprehensive coverage of 10-gigabit optical networks in provincial industrial parks [2]

党的二十届四中全会提出，"全面实施'人工智能+'行动"。"四川数据资源丰富、产业体系完备、 应用场景广阔，加快推动人工智能赋能新型工业化大有可为。"省经济和信息化厅相关负责人介 绍。2024年，全省人工智能产业营收超1300亿元，同比增长超25%，智能化发展底座坚实。 行动方案明确了四川推动人工智能赋能新型工业化的阶段性目标。具体而言，到2027年，全省规 模以上工业企业实现数字化转型全覆盖，打造50个以上数字产业园区，培育90家以上人工智能领 域国家制造业单项冠军企业、专精特新"小巨人"企业等。 行动方案提出5大重点任务，涉及强化科技研发创新、产品迭代升级、场景应用牵引、治理机制构 建、产业生态打造等方面，并细分为18项子任务。 ●到2027年，全省规模以上工业企业实现数字化转型全覆盖，打造50个以上数字产业园区，培育 90家以上人工智能领域国家制造业单项冠军企业、专精特新"小巨人"企业等 比如，在强化智能化技术科技创新上，四川将加快构建人工智能全链条自主创新体系，每年重点 支持高水平智能软件、硬件技术攻关，按照项目总投资，给予最高50%、不超过1000万元资金支 持。重点攻关的方向，则包括预训练大模型、高 ...

Core Insights - The Pentagon awarded Alphabet a significant defense contract to lead the GenAI.mil platform, which is part of the White House's AI Action Plan [2][3] - This move is notable as Palantir Technologies has been a dominant player in defense operations, particularly in data mining and AI applications [1][3] Group 1: GenAI.mil Platform - GenAI.mil is designed to provide government workers with a suite of generative AI tools, enhancing capabilities in natural language processing and agentic workflows [6] - The platform is built on Google's AI system, Gemini, which aims to facilitate rapid software acquisition and implementation within the Pentagon [5][12] Group 2: Competitive Landscape - Historically, defense contracts have favored companies like Boeing, Lockheed Martin, and Palantir, making Alphabet's selection for GenAI.mil a surprising development [8] - Alphabet previously led Project Maven, a military initiative for machine learning, but faced employee protests that led to the non-renewal of the contract [9][11] Group 3: Implications for Palantir - Palantir has established itself with tools like Foundry, Gotham, and Apollo, which are critical for defense operations [13][15] - Despite Alphabet's win, the coexistence of Palantir's AI Platform and Alphabet's Gemini is anticipated, suggesting that both companies can thrive in the defense tech space [16] - The government's investment in AI across multiple high-performance players indicates a broader commitment to enhancing defense capabilities [17]

Core Viewpoint - The artificial intelligence (AI) revolution is described as a significant investment opportunity, with the potential to transform various industries [1] Group 1: Investment Opportunities in AI - Investing in AI presents challenges due to the vast number of companies, making it difficult to identify which will succeed [2] - An exchange-traded fund (ETF) focused on AI provides diversified access to the industry, with the Global X Artificial Intelligence & Technology ETF (AIQ) being highlighted as a strong option [3] Group 2: ETF Characteristics - The Global X AIQ ETF includes well-known companies such as Alphabet, Broadcom, Nvidia, and Palantir, similar to other AI-focused ETFs [5] - AIQ is distinguished by its balanced assembly, mirroring the Indxx Artificial Intelligence & Big Data index, which categorizes holdings into AI developers and service providers, and AI hardware [6][7] - The ETF employs a unique weighting methodology, limiting companies with significant AI exposure to a maximum of 3% of the index's total value, and those with modest exposure to 1% [9] Group 3: Performance and Strategy - The allocation approach of AIQ has shown advantages, particularly since April, amidst concerns of an AI bubble [11] - Holding the Global X AIQ ETF allows for balanced exposure to the AI sector, mitigating risks associated with volatility and profit-taking from a few high-performing stocks [13][14]

Core Insights - The article highlights the rapid success story of Xiao Hong, who transitioned from an entrepreneur to a billionaire in just one and a half years by selling his company Manus to Meta, marking a significant achievement in the AI industry [1][3][12] Company Overview - Manus, founded by Xiao Hong, is recognized as the world's first general-purpose intelligent agent, designed not for conversation but for action [8] - The company achieved an annual recurring revenue (ARR) of over $100 million within eight months, showcasing an unprecedented growth rate in business history [10] Industry Implications - Xiao Hong's success illustrates a shift in the Chinese AI landscape, moving towards global markets and emphasizing the importance of practical applications over theoretical models [12][15] - The acquisition by Meta signifies a strategic move to access the shortest path to AGI (Artificial General Intelligence) applications, positioning Manus as a key player in this transition [11] Entrepreneurial Lessons - The story serves as a lesson for Chinese entrepreneurs, emphasizing that the AI era offers a shorter window for capitalizing on opportunities, with the potential for rapid wealth creation [15] - The focus on providing actionable tools rather than just software solutions reflects a new business model in the tech industry, where efficiency and user-centric design are paramount [9][13]

Core Viewpoint - Elon Musk's AI image generator, Grok, is facing scrutiny for generating nonconsensual sexualized images, including those of minors, raising ethical concerns and prompting investigations by authorities [1][2][4]. Group 1: Ethical Concerns and Investigations - Grok has been exploited by users to create fake images of individuals in revealing outfits, leading to significant backlash and investigations by French authorities [1][2]. - India's Ministry of Electronics and Information Technology has called for a comprehensive review of Grok and the removal of content violating Indian laws [2]. - The UK's Minister for Victims & Violence Against Women and Girls has urged Musk to address the exploitation of women through AI-generated images [2]. Group 2: Company Response and Safeguards - In response to the backlash, Grok acknowledged "lapses in safeguards" and stated that urgent fixes are being implemented, though it is unclear if this response was reviewed by xAI or generated by AI [3]. - The incident highlights the ongoing challenge of deepfakes for AI companies and emphasizes the need for stricter regulations and safeguards to prevent misuse [4]. Group 3: Reputational Risks - The backlash against Grok underscores the potential reputational risks for companies when their products are used unethically, serving as a reminder for AI companies to prioritize user safety and privacy [5].

Core Insights - The article discusses the emergence of AI video generation technology, highlighting the journey of Wang Changhu and his company, Aishi Technology, which aims to democratize video creation for everyone [1][9][32] Group 1: Industry Overview - The AI industry is experiencing a significant transformation, comparable to the impact of steam engines, electricity, and computers, with AI models becoming a universal technology [1][4] - 2022 marked the beginning of the AIGC era, with tools like Midjourney and ChatGPT showcasing the potential of AI, leading to a new wave of interest and investment in AI technologies [4][5] - The AI video generation sector was initially viewed skeptically by experts, who doubted its viability within five years, yet Aishi Technology has successfully launched a leading AI video model [6][10] Group 2: Company Strategy - Aishi Technology's strategy focuses on video generation, a less popular but potentially more impactful area compared to text and image models, allowing the company to carve out a unique niche [10][11] - The company aims to make video creation accessible to everyone, transforming ordinary users into directors of their own content, thus lowering the barriers to entry in video production [13][14] - The product "拍我AI" (Shoot Me AI) exemplifies this philosophy, enabling users to create videos with simple prompts, significantly enhancing user engagement and creativity [15][16] Group 3: Growth Mechanism - Aishi Technology has developed a self-reinforcing growth cycle, where high-quality models attract users, who in turn provide valuable data that enhances the models and products [20][22] - The company’s approach combines advanced technology with user-friendly design, resulting in a product that appeals to a broad audience and encourages content creation [21][22] Group 4: Organizational Efficiency - Aishi Technology emphasizes extreme organizational efficiency and precise technical judgment to compete with larger firms, aiming for ten times the efficiency of its competitors [24][25] - The company adopts a flat organizational structure to facilitate rapid decision-making and resource allocation, enhancing overall productivity [26][27] - Continuous innovation and attracting top talent are crucial for maintaining a competitive edge in the rapidly evolving AI landscape [28]

Core Viewpoint - MiniMax's IPO journey reflects a dual challenge of rapid C-end market growth and significant financial losses, copyright disputes, and intense industry competition, marking a critical phase in the commercialization of large models in China [1] Group 1: Financial Performance - MiniMax's revenue skyrocketed from $3.46 million in 2023 to $30.52 million in 2024, representing a staggering year-on-year growth of 782.2% [3] - In the first three quarters of 2025, revenue surpassed $53.44 million, exceeding the total revenue of 2024 by 174.7% [3] - The company's core product, Talkie, contributed over 70% of its revenue, with international markets accounting for more than 70% of total income, reaching over 212 million users globally [3] Group 2: Product and Market Dynamics - Talkie, with its highly customizable virtual characters, has successfully addressed users' emotional companionship needs, topping the global free chart for AI companion dialogue products [5] - The Hailuo-02 video generation model has rapidly gained traction, increasing its revenue contribution from 7.7% in 2024 to 32.6% in the first three quarters of 2025, with an ARPU of $56, significantly higher than Talkie's [5] - Monthly active users of AI-native products surged from 3.1 million in 2023 to 27.6 million in the first three quarters of 2025, with paid user numbers exceeding 1.77 million [5] Group 3: Financial Challenges - MiniMax faced a cumulative net loss of $12.5 billion from 2022 to the first three quarters of 2025, with R&D expenses reaching $180 million in the first three quarters of 2025, maintaining a high R&D expense ratio of 337.4% [8] - Marketing expenses peaked at $86.99 million in 2024, with a significant portion of $39.32 million in the first three quarters of 2025, accounting for 10.48% of revenue [8] - The gross margin for AI-native products was -8.1% in 2024, only turning positive at 4.7% in 2025, indicating a fragile profitability foundation [10] Group 4: Legal and Competitive Landscape - MiniMax is embroiled in copyright disputes, facing a lawsuit from Disney, Universal Pictures, and Warner Bros. for $75 million over unauthorized use of IPs for training models [12] - The company has been accused of illegal training data usage and inducing infringement, with previous lawsuits from iQIYI marking significant legal challenges [12] - The competitive landscape is intensifying, with major players like OpenAI and Character.AI dominating technology, while domestic giants are launching aggressive strategies against MiniMax [16]