21世纪经济报道记者周慧 北京报道 "十年来，长江经济带优良水质比例从67%提升到96.5%，地区生产总值翻了一番多、占全国比重从42.2%提升到47.3%，实现了保护和发展的相互促进、相 得益彰。"王昌林表示。 图：发布会现场 拍摄：周慧 王昌林还表示，十年来，横贯东西、承接南北、通江达海的优势在长江经济带持续释放，开放长江活力更加澎湃。长江水系高等级航道由0.8万公里增加到 1.1万公里，干线港口货物吞吐量增长71%、达到42亿吨、稳居世界内河第一。推进高水平对外开放，中欧（亚）班列、西部陆海新通道高质量建设，沿江9 个自贸试验区联动发展。 1月5日，国新办举行新闻发布会介绍长江经济带发展十年工作进展和成效，国家发展改革委副主任王昌林表示，十年来，科技创新和产业创新在长江经济带 加快融合，创新长江动能更加强劲。 "我们坚持把科技创新作为主动力，持续推进上海（长三角）国际科技创新中心和武汉、成渝地区区域科技创新中心建设，积极培育壮大新兴产业和未来产 业。"王昌林表示，长江经济带创造出"人造太阳"亿度千秒运行、靶向抗癌药等一批处于世界先进水平的科技成果，涌现出DeepSeek、宇树科技等一批具有 国际竞争力的 ...

Satellite Internet - Blue Arrow Aerospace's IPO on the Sci-Tech Innovation Board has been accepted, aiming to raise 7.5 billion yuan, with a focus on reusable rockets [1] - Starlink announced it will provide free broadband service to the people of Venezuela for one month [3] Artificial Intelligence - Wall Street is seeing active IPOs and financing, with companies like Biran Technology experiencing a significant stock price increase, and Suiyuan Technology completing IPO guidance [4] - Google and Anthropic reported strong demand for TPU, while there are rumors of significant price increases for NVIDIA and AMD GPUs starting next month [4] Semiconductor - The Big Fund has increased its stake in SMIC H-shares, and TrendForce has revised upward its expectations for DRAM and NAND contract price increases in Q1 2026 [5] - TSMC has received U.S. permission to transport equipment to its Nanjing plant and plans to raise prices for advanced processes [6] Robotics - Reports suggest that Tesla's Optimus V3 mass production review has been completed, with seven Chinese companies as core suppliers, targeting a production capacity of 50,000 to 100,000 units by the end of the year [7] New Energy - Tianci Materials has released an annual forecast, confirming that Q4 performance is on track [9] Nuclear Fusion - The all-superconducting tokamak experiment has achieved significant physical breakthroughs, providing a basis for high-density operation [10] - The first nuclear fusion conference of the year will be held from January 16 to 17 in Hefei [11] Autonomous Driving - Tesla owners have achieved "full self-driving across the United States" for the first time [13] Digital Currency - State-owned banks will start paying interest on digital RMB real-name wallet balances at current account rates from January 1 [14] Market News - The New Year speech mentioned advancements in artificial intelligence, semiconductors, aerospace, and robotics across multiple sectors [15] - Companies are required to disclose annual performance forecasts by January 31 [16] - Market pricing clues for 2026 indicate sustained high growth, broad recovery, U.S. inflation, and Chinese domestic demand [17] Market Observation - On the last trading day before the holiday, the transaction volume reached 2.05 trillion yuan, with indices closing in the green but not reaching 4,000 points [18] - The satellite internet sector successfully crossed the year-end, with funds remaining in the market, and there is potential for healthy pullbacks to provide better entry opportunities [18]

Core Insights - The interview with Professor Zhou Qiren from Peking University highlights his unique approach to economics, emphasizing practical business experiences over abstract theories [1][2] - Zhou believes that companies must navigate their own paths in uncertain environments, advocating for a focus on customer needs rather than merely competing with rivals [4][10] Group 1: Research Methodology - Zhou emphasizes the importance of real-world research, stating that understanding comes from engaging with actual business practices rather than theoretical frameworks [2][4] - He encourages a mindset of inquiry and debate, suggesting that true insights emerge from discussions and challenges to prevailing views [1][4] Group 2: Business Strategies - Zhou's new book, "Seeking Path Collection," discusses various strategies for companies to thrive amidst uncertainty, advising them to adapt based on their specific circumstances [4][6] - He highlights the significance of understanding customer needs and positioning, arguing that many companies fail because they do not prioritize customer insights [10][15] Group 3: Market Competition - Zhou views the current competitive environment, referred to as "卷" (competition), as a double-edged sword, suggesting that while it can drive prices down, it can also lead to unhealthy market practices [4][5] - He warns against focusing solely on competitors, advocating for a customer-centric approach that prioritizes service quality over speed [11][15] Group 4: Global Market Positioning - Zhou argues that companies should not rigidly separate domestic and international markets but should instead seek opportunities based on customer demand globally [19][20] - He notes that many businesses can replicate successful strategies from their home markets in international contexts, emphasizing the need for adaptability [21][20] Group 5: Innovation and Government Role - Zhou critiques the notion that innovation can be mandated by government policies, suggesting that true innovation arises from grassroots efforts and real-world applications [22][24] - He believes that local governments should focus on their unique strengths rather than trying to replicate successful models from other regions [22][25]

AI Regulation and Governance - Elon Musk's xAI has introduced a new image editing feature for Grok, allowing users to edit images without the original author's consent, leading to controversy and misuse [1] - xAI acknowledged the flaws in its protective measures and is working on urgent fixes, emphasizing that content involving minors is illegal and prohibited [1] AI Safety and Recruitment - OpenAI is hiring a new Head of Preparedness with an annual salary of $555,000 (approximately 3.89 million RMB) to develop safety processes for AI models [2] - The role is crucial as AI capabilities are rapidly advancing, posing real-world challenges, including potential impacts on mental health and cybersecurity [2][3] - OpenAI aims to create effective constraints to minimize risks while maximizing the benefits of AI [3] AI Applications and Developments - Ant Group's AI health app, "Ant Aifu," clarified that its Q&A results contain no advertisements or commercial rankings, focusing on health management [4] - The app has over 15 million monthly active users, ranking among the top five AI apps in China, with 55% of users from lower-tier cities [4] AI Job Market Impact - AI expert Geoffrey Hinton predicts that AI will replace more jobs in the coming year, with software engineering projects requiring minimal human involvement [5][6] - Hinton expressed concerns about the societal risks of AI, criticizing insufficient responses to these challenges [6] AI Technology Testing and Partnerships - Apple is reportedly testing an AI feature for its iPhone in China, but no official announcement has been made yet [7][8] - Meta has acquired the AI startup Manus for several billion dollars, marking its third-largest acquisition to date [9] - The acquisition is part of Meta's strategy to enhance its superintelligence capabilities [9] IPO and Market Movements - Zhiyuan Huazhang has initiated a global IPO, aiming to be the first publicly listed company focused on general AI models in Hong Kong [10][11] - The company plans to issue approximately 37.42 million shares at a price of HKD 116.20 per share, potentially raising HKD 4.3 billion [11] Research and Innovations - DeepSeek has published a paper indicating that its new model, DeepSeek V4, has completed training, with expectations for its release around the Lunar New Year [12] - Volcano Engine has become the exclusive AI cloud partner for the 2026 Spring Festival Gala, leveraging advanced AI technologies for the event [13] New Hardware Developments - OpenAI is collaborating with former Apple design chief Jony Ive on a hardware project, codenamed Gumdrop, with multiple products in development [14] - The new hardware aims to create an ecosystem of devices, with potential releases expected in 2026 [14]

Investment Rating - The industry investment rating is maintained as "Recommended" [2] Core Insights - The report highlights a divergence in technology market performance between US and Hong Kong stocks during the holiday period, with AI and commercial aerospace sectors showing frequent information catalysts, suggesting continued focus on core industry players [1][6] - The report emphasizes the importance of AI applications and the acceleration of large model iterations, recommending continued attention to companies like Alibaba, Kingsoft, and others in the AI infrastructure space [6] Summary by Sections Holiday Technology Dynamics - During the New Year holiday, US and Hong Kong stock markets exhibited divergent trends, with technology stocks being the main market drivers. Notably, AI and storage chips led the gains, with Micron Technology rising over 10% to reach a historical high, and significant increases in other tech stocks [9] - The Hong Kong market saw a "New Year rally," with the Hang Seng Index rising 2.76% and the Hang Seng Technology Index increasing by 4% [9] Market Performance Review - In the last week of 2025, the computer sector rose by 1.16%, with the top-performing stocks being Yuyin Co., Digital Certification, and others, while the report also lists the top gainers and losers in the market [24]

Group 1 - The term "slaughter line" has gained popularity on social media, highlighting the harsh reality of financial crises leading to homelessness in the U.S. [1][2][4] - Many Silicon Valley engineers have faced bankruptcy due to unexpected events like job loss or illness, with some high-earning professionals becoming homeless within months [2][4] Group 2 - Xiaomi's CEO Lei Jun conducted a live stream where he disassembled a new Xiaomi YU7 to address various online concerns about the product [7][8] - The live stream was initially delayed due to Lei Jun's illness but became significant as it was the first major live event of 2026 [7] Group 3 - Lenovo's CEO Yang Yuanqing and NVIDIA's CEO Jensen Huang discussed a new collaboration focused on "hybrid AI" during a conversation ahead of CES 2026 [8][9] - They identified two key trends in AI development: enterprise-level AI as a core battleground and the integration of AI into various sectors of the economy [8][9] Group 4 - Baidu plans to spin off its AI chip subsidiary Kunlun and has submitted a listing application to the Hong Kong Stock Exchange, with an estimated valuation of at least $3 billion [13][15] - The spin-off aims to enhance Kunlun's market image and attract investors focused on AI computing chips and related systems [15] Group 5 - A new regulation on green product certification in China has been introduced, marking a significant shift towards comprehensive regulation of the certification process [16] - The updated guidelines cover 122 product categories, including electronics and packaging, and aim to standardize the certification process [16] Group 6 - The company "Moon's Dark Side" has completed a $500 million Series C funding round, raising its valuation to $4.3 billion [17][18] - The funding will be used to expand graphics card production and accelerate the training and development of its K3 model [17] Group 7 - OpenAI is at a critical juncture in 2026, needing to demonstrate a sustainable business model after years of rapid growth and significant funding [20] - Investors are shifting from a phase of leniency to demanding clear profitability pathways from OpenAI [20] Group 8 - Cloud Leopard Intelligent has initiated its IPO process, aiming to become the first domestic DPU (Data Processing Unit) stock in China [30][33] - The company specializes in cloud computing and DPU solutions, with a valuation exceeding 14 billion yuan as of late 2025 [33] Group 9 - Meta has acquired the AI application company Manus for several billion dollars, marking its third-largest acquisition to date [33][34] - Manus, which was previously valued at $2 billion, is known for its AI agent product that automates complex tasks based on simple user instructions [34][35]

Investment Rating - The industry is rated as "Overweight," indicating an expected increase in value exceeding 10% compared to the market benchmark index over the next 6-12 months [6]. Core Insights - The large model has entered a post-training and COT expansion phase, with the capabilities of the base model in 2025 likely remaining unchanged, as the GPT-5 series may still utilize the GPT-4o base model. The focus for 2025 will be on enhancing post-training and reasoning capabilities [1]. - Google's Gemini 3 model has achieved significant advancements in cross-modal dialogue, understanding, and content generation, but still faces challenges with logical coherence in complex scenarios and controllability of generated content, highlighting key areas for future technological breakthroughs [1]. - Domestic AI chip manufacturers have reached H-series performance levels, with advancements in interconnect speeds and software ecosystem capabilities. Notably, Alibaba's latest PPU chip has surpassed NVIDIA's A800 in key performance metrics, and Huawei's CloudMatrix 384 super node aims to optimize computing efficiency [2]. - The capabilities of robotic bodies have improved significantly, while the cognitive abilities of their "brains" lag behind, limiting their application to structured scenarios. The VLA model architecture has been criticized for its limitations in real-time reasoning in complex physical environments [3]. - The business models for AI applications are still under exploration, with domestic companies facing challenges in monetization despite high revenue growth rates, while international firms struggle with high computing costs and low profit margins [3].

Core Insights - DeepSeek has introduced a new neural network architecture optimization called mHC (Manifold-Constrained Hyper-Connections), which is expected to significantly impact the AI industry, including large models and chips [1][5][9] Group 1: mHC Architecture - The mHC architecture builds on the Hyper-Connections (HC) framework released by the Byte Bean team in November 2024, aiming to replace the nearly decade-old ResNet architecture [5] - mHC introduces a Manifold-Constrained approach using the Sinkhorn-Knopp algorithm to stabilize signal propagation during training, addressing issues of signal explosion and instability in large model training [5][6] - In training demonstrations with 27 billion parameters, mHC maintained a signal amplification of only 1.6 times, while HC experienced a catastrophic failure with a 3000 times amplification [6][8] Group 2: Performance and Efficiency - mHC shows a significant reduction in training loss and improved performance on challenging tasks, with over 2% enhancement in reasoning and reading comprehension benchmarks compared to traditional architectures [6][8] - The additional training time overhead for mHC, even with a fourfold expansion of residual channels, is only 6.7%, indicating a focus on cost-effectiveness and efficiency [8] Group 3: Industry Impact and Reactions - The release of mHC has sparked high discussion levels among researchers and industry professionals, with expectations of a paradigm shift in large model architectures by 2026 [9][10] - Competitors are already responding, with new architectures like Deep Delta Learning emerging shortly after mHC's announcement, indicating a potential chain reaction in AI architecture development [9][10] - Analysts predict that DeepSeek may make significant announcements around the Lunar New Year, potentially unveiling the long-awaited R2 model or a faster universal model V4 [10] Group 4: Compatibility and Market Dynamics - mHC's architecture is primarily designed for NVIDIA's supernode links, raising concerns about compatibility with domestic chips, which may require enhanced adaptation efforts [11] - As U.S. AI chip manufacturers gradually exit the Chinese market due to geopolitical factors, domestic chipmakers are accelerating their development and ecosystem building to adapt to DeepSeek's models [12]

Group 1: Meta's Acquisition of Manus - Meta has acquired the AI startup Manus for a price reported to be in the billions, marking it as Meta's third-largest acquisition after WhatsApp and Scale.ai [1] - Manus will continue its operations in Singapore and maintain its product offerings through its app and website without changes to its decision-making processes [1] - The acquisition reflects Meta's urgency to enhance its AI capabilities, especially in light of competition from Google's Gemini 3 [1] Group 2: SoftBank's Investment in OpenAI - SoftBank has completed a $40 billion investment commitment to OpenAI, making it one of the largest private financings in history [2] - The final tranche of the investment, amounting to $22 billion to $22.5 billion, has been sent recently [2] - SoftBank's divestment of Nvidia shares for $5.83 billion indicates a strategic shift to fund AI projects, including the partnership with OpenAI [2] Group 3: DeepSeek's New Research - DeepSeek has introduced a new network architecture called mHC (manifold-constrained hyperconnection) aimed at improving model training stability and efficiency [3] - The research addresses issues related to the scalability and memory access costs of existing hyperconnection models [3] - Industry experts view this innovation as a foundational advancement that could lead to significant updates in future versions of DeepSeek's technology [3] Group 4: Moonlight's Financing and Market Position - Moonlight, a large model unicorn, has completed a $500 million Series C financing, significantly exceeding its target, and currently holds over 10 billion yuan in cash [4] - The funds will be used to aggressively expand GPU resources and accelerate the training and development of its K3 model [4] - Moonlight aims to surpass competitors like Anthropic to become a leading AGI company [4] Group 5: Upcoming IPOs in the AI Sector - Companies including OpenAI, Anthropic, and SpaceX are preparing for potential IPOs this year, with total fundraising expected to reach hundreds of billions [6] - OpenAI is negotiating a new valuation of $750 billion, while Anthropic's valuation may exceed $300 billion [6] - The combined valuation of these companies could approach 13 trillion yuan, indicating a significant market impact [6] Group 6: MiniMax's IPO Plans - MiniMax has initiated its IPO process, aiming to raise up to 4.19 billion HKD (approximately $538 million) with a share price range of 151 to 165 HKD [7] - The company is set to list on the Hong Kong Stock Exchange on January 9, 2026, shortly after its competitor, Zhipu AI [7] - MiniMax's cornerstone investors include major financial institutions and investment funds, highlighting strong market interest [7] Group 7: Baidu's Kunlun Chip IPO - Baidu has submitted a confidential application for its AI chip subsidiary Kunlun to independently list on the Hong Kong Stock Exchange [8] - This move follows Baidu's earlier evaluation of the potential for a spin-off, indicating a strategic shift in its business model [8] - The competitive landscape for Kunlun includes major players like Nvidia and AMD, as well as domestic competitors [8] Group 8: Wall Street's Response to Wall Street's IPOs - Wall Street analysts predict that if any of the aforementioned companies successfully go public, it could overshadow the total fundraising of approximately 200 companies in the U.S. in 2025 [6] - The anticipated IPOs are expected to generate significant returns for venture capitalists and investment bankers involved in the transactions [6] Group 9: Wall Street's Response to Wall Street's IPOs - Wall Street analysts predict that if any of the aforementioned companies successfully go public, it could overshadow the total fundraising of approximately 200 companies in the U.S. in 2025 [6] - The anticipated IPOs are expected to generate significant returns for venture capitalists and investment bankers involved in the transactions [6] Group 10: xAI's Expansion - xAI, led by Elon Musk, has purchased a third building to enhance its training capabilities, aiming for nearly 2 gigawatts of computing power [15] - The new facility is set to be transformed into a data center by 2026, supporting xAI's growth and operational needs [15] - xAI's previous investments in data centers indicate a strong commitment to expanding its AI infrastructure [15]

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 ...