Core Insights - The article highlights the explosive growth of the satellite and AI application sectors, marking the beginning of a strong market trend in 2026, with A-shares achieving a 17-day consecutive rise and nearing 4200 points [1][24]. Group 1: Satellite Industry Developments - China has submitted a record application to the International Telecommunication Union (ITU) for approximately 203,000 satellites, covering 14 constellations, aiming to secure essential resources for satellite internet development over the next decade [6][26]. - The U.S. Federal Communications Commission (FCC) has approved SpaceX's deployment of an additional 7,500 second-generation Starlink satellites, indicating a competitive race for scarce space resources [7][8]. - The satellite industry is experiencing a significant increase in investment, with a notable rise in institutional fund inflows exceeding 150% in key sectors such as satellite manufacturing and rocket support [16][17]. Group 2: AI Application Advancements - The AI sector is witnessing a shift from "Chat" to "Agent" models, with major tech companies accelerating their entry into AI healthcare and other applications [11][12]. - The cost of AI services has decreased by over 90% in the past 18 months, making complex AI applications economically viable for large-scale commercialization [25][28]. - Significant investments are being made in AI, with SoftBank investing $41 billion in OpenAI and ByteDance partnering with NVIDIA for a $14 billion chip collaboration [20][22]. Group 3: Market Trends and Investment Opportunities - The satellite ETF E Fund (563530) has seen a net inflow of 860 million yuan, reflecting strong investor interest in the satellite industry, which is expected to benefit from a long-term growth cycle [18][20]. - The software ETF E Fund (562930) has increased by 213% in size compared to the previous year, indicating a robust interest in AI application companies [20][22]. - The convergence of satellite technology and AI applications is anticipated to create substantial commercial value, with both sectors poised for significant growth in 2026 [30][31].

Core Insights - The article discusses the evolution of artificial intelligence (AI) in 2025, highlighting a shift from merely increasing model parameters to enhancing model intelligence through foundational research in areas like fluid reasoning, long-term memory, spatial intelligence, and meta-learning [2][4]. Group 1: Technological Advancements - In 2025, significant technological progress was observed in fluid reasoning, long-term memory, spatial intelligence, and meta-learning, driven by the diminishing returns of scaling laws in AI models [2][3]. - The bottleneck in current AI technology lies in the need for models to not only possess knowledge but also to think and remember effectively, revealing a significant imbalance in AI capabilities [2][4]. - The introduction of Test-Time Compute revolutionized reasoning capabilities, allowing AI to engage in deeper, more thoughtful processing during inference [6][10]. Group 2: Memory and Learning Enhancements - The Titans architecture and Nested Learning emerged as breakthroughs in memory capabilities, enabling models to update their parameters in real-time during inference, thus overcoming the limitations of traditional transformer models [19][21]. - Memory can be categorized into three types: context as memory, RAG-processed context as memory, and internalized memory through parameter integration, with significant advancements in RAG and parameter adjustment methods [19][27]. - The introduction of sparse memory fine-tuning and on-policy distillation methods has mitigated the issue of catastrophic forgetting, allowing models to retain old knowledge while integrating new information [31][33]. Group 3: Spatial Intelligence and World Models - The development of spatial intelligence and world models was marked by advancements in video generation models, such as Genie 3, which demonstrated improved physical understanding and consistency in generated environments [35][36]. - The emergence of the World Labs initiative, led by Stanford professor Fei-Fei Li, focused on generating 3D environments based on multimodal inputs, showcasing a more structured approach to AI-generated content [44][46]. - The V-JEPA 2 model introduced by Meta emphasized predictive learning, allowing models to grasp physical rules through prediction rather than mere observation, enhancing their understanding of causal relationships [50][51]. Group 4: Reinforcement Learning Innovations - Reinforcement learning (RL) saw significant advancements with the rise of verifiable rewards and sparse reward metrics, leading to improved performance in areas like mathematics and coding [11][12]. - The GPRO algorithm gained popularity, simplifying the RL process by eliminating the need for a critic model, thus reducing computational costs while maintaining effectiveness [15][16]. - The exploration of RL's limitations revealed a ceiling effect, indicating that while RL can enhance existing model capabilities, further breakthroughs will require innovations in foundational models or algorithm architectures [17][18].

1月12日，A股市场震荡拉升，三大指数均涨超1%。沪深两市成交额3.6万亿创历史新高，连续第2个交 易日突破3万亿。板块方面，AI应用再度大爆发，Sora概念股集体狂欢，流金科技30cm涨停，福石股 份、拓尔思、易点天下、值得买、万兴科技等多股20cm涨停。 消息面上，据报道，DeepSeek计划在未来几周内发布新一代AI模型V4，预计发布时间在2月中旬（春 节）前后，但具体时间仍可能调整。报道称，根据DeepSeek内部基准测试结果，V4在编程任务上的表 现超越了目前市场上的主流模型。 平安证券认为，全球AI算力基础设施能力的持续提高，与全球大模型的持续迭代升级共振，将推动全 球AI产业持续蓬勃发展。当前，智谱、MINIMAX等我国大模型算法公司陆续登录资本市场，我国大模 型商业化的进程有望进一步加快。坚定看好我国AI产业的发展前景，建议持续关注AI主题的投资机 会，关注AI算力、AI算法和应用两个方向的投资机会。 （文章来源：21世纪经济报道） ...

Core Viewpoint - The AI application sector has experienced a significant surge, with multiple stocks reaching their daily limit up, indicating strong market interest and potential investment opportunities in this area [1][8]. Group 1: Market Performance - Five stocks, including Puyuan Information and New Point Software, hit a 20% limit up, while three others, such as Yonyou Network, achieved a 10% limit up [1][9]. - The Innovation ETF (562030), heavily invested in the software development sector, saw its price rise by 7%, indicating strong market momentum [9]. Group 2: Reasons for AI Application Surge - The recent strong market performance of domestic large model companies MiniMax and Zhipu, which have listed on the Hong Kong stock market, signals the successful establishment of a "capital closed loop" for domestic large models, marking the beginning of the "commercial value realization" of AI applications [3][10]. - Three new hotspots in AI applications have emerged: AI for Science (AI4S), Generative Engine Optimization (GEO), and AI in healthcare, with significant user engagement reported for new AI health features [3][10]. - New advancements in large model technology include the release of Qwen3-VL-Embedding and Qwen3-VL-Reranker models, as well as a new architecture proposed by DeepSeek [3][10]. Group 3: Investment Recommendations - Key investment directions in AI applications include: 1. Super entrance: Large models have evolved into dominant traffic entry points in the AI era [4][11]. 2. AI Infrastructure: Software-defined computing is expected to yield stable returns [4][11]. 3. High growth: AI technology is driving commercialization in marketing and entertainment sectors [4][11]. 4. High barriers: Data flow and workflow integration in sectors like healthcare and manufacturing present significant opportunities [4][11]. Group 4: Industry Outlook - The domestic innovation industry is expected to see a recovery in sentiment due to large-scale debt measures and proactive fiscal policies [4][11]. - The current geopolitical climate and the push for self-sufficiency in technology highlight the urgent need for development in the innovation sector, supported by government initiatives [13].

Group 1 - AI concept stocks saw significant gains, with Zhihui (02513) up 61.41% to HKD 256, MINIMAX (00100) up 29.91% to HKD 448.2, and Maifushi (02556) up 26.08% to HKD 48.92 [1] - DeepSeek is set to release its next-generation flagship model V4 in the coming weeks, which reportedly outperforms mainstream models like Claude and ChatGPT in code generation capabilities [1] - CITIC Securities believes that the current dynamics in the AI industry, including financing activities from overseas companies like xAI and Anthropic, along with domestic policies promoting "AI + manufacturing," will lead to a new wave of AI application enthusiasm following the launch of DeepSeek-V4 [1] Group 2 - Zhihui's stock surged over 60%, leading the AI sector, with its market capitalization surpassing HKD 100 billion [2] - The company announced a strategic partnership with Didi to explore key technologies in general artificial intelligence (AGI) and their applications in the transportation sector [2] - Dongwu Securities expressed optimism about Zhihui's strengths in local model technology, open-source ecosystem, and localized implementation capabilities, anticipating benefits from the long-term trend of transitioning from localized deployment to cloud services in China's large model industry [2]

招商证券表示，建议投资者围绕产业趋势最为明确的AI、政策驱动最为确定的信创以及最为受益"牛 市"Beta 的金融科技坚定布局。资料显示，云计算ETF华夏（516630）跟踪云计算指数（930851），费 率最低。该指数聚焦国产AI软硬件算力，算机软件+云服务+计算机设备合计权重高达83.7%，deep seek、AI应用含量均超40%。场外联接（A类：019868；C类：019869） 1月12日，AI+方向集体沸腾，截至13:35，低费率云计算ETF华夏(516630)上涨6.47%， 冲击3连涨，持 仓股拓尔思、汉得信息、易点天下20cm涨停，万兴科技，中科星图等个股跟涨。拉长时间看，截至 2026年1月9日，云计算ETF华夏近1周累计上涨9.33%。 消息面上，DeepSeek将于2月发布新一代旗舰AI模型DeepSeek V4，该模型具备强大的编程能力，预计 将对当前的AI竞争格局产生重大影响。V4是DeepSeek继2024年12月发布的V3模型之后的最新版本。知 情人士称，DeepSeek内部的初步测试表明，V4在编程能力上超过了目前市场上的其他顶级模型，如 Anthropic的Claude和Op ...

Group 1 - The Hong Kong stock market saw a strong performance on January 12, with significant gains in tech, media, and aerospace sectors, particularly AI concept stocks [1] - Notable performers included Zhihui, which surged nearly 38%, and MINIMAX, which rose over 25% [1] - Other companies that experienced substantial gains include Lion Group (up over 28%), Kingsoft Cloud (up 13%), Meitu (up 11%), and several others in the tech sector [1][2] Group 2 - Bloomberg industry research indicates that the earnings growth of Chinese tech giants is expected to surpass that of the "seven giants" in the US stock market for the first time since 2022, with a significant turning point anticipated in 2026 [3] - DeepSeek is set to release its next-generation V4 model in mid-February, which is expected to excel in programming tasks, long code prompt processing, and improved data pattern understanding [3] - The development of large models is accelerating the commercialization of AI applications, particularly in search & marketing, coding, multimodal applications, agents, and AI for science [3] - The evolution of large models is seen as a key tool for companies with deep industry know-how, particularly in AI healthcare, which is transitioning from a single auxiliary tool to comprehensive lifecycle management [3]

Group 1 - The core viewpoint of the news highlights the strong performance of the Sci-Tech Innovation Artificial Intelligence ETF (589010), which surged by 4.33%, indicating robust market sentiment towards AI investments [1][2] - The ETF's trading volume exceeded 252 million yuan, with a turnover rate of over 8%, reflecting high trading enthusiasm and recognition of long-term value in the AI sector [1] - Key constituent stocks such as New Point Software and Hai Tian Rui Sheng experienced significant gains, with New Point Software hitting a 20% limit up and several others rising over 15%, showcasing excellent profit potential [1] Group 2 - Open Source Securities noted that AI innovation is continuously evolving, with model capabilities improving and costs decreasing, particularly with the rise of Chinese open-source models like DeepSeek and Qwen [2] - The development of multi-modal large models is experiencing rapid breakthroughs, which is expected to further enhance application growth in the AI industry [2] - The Sci-Tech Innovation Artificial Intelligence ETF closely tracks the Shanghai Stock Exchange Sci-Tech Innovation Board AI Index, covering high-quality enterprises across the entire industry chain, benefiting from high R&D investment and policy support [2]

大模型不赚钱，这事不多说，但简单回顾： OpenAI据说得2029年盈利，每年亏140亿刀。 Anthropic据说2028年能盈利，也亏损。 刚上市的MiniMax每年亏5个亿刀。 智谱和MiniMax差不多，也是每年亏这么多。 据说DeepSeek能赚钱，但这行当基础模型基本亏损是确定的。 不能赚钱就主要靠VC，国内VC不给力所以就导致了基模发展的一系列问题。 但基模属于非典型产品，不多说了，下面重点说智能体。（包括做基础模型也做应用的） 智能体也先说结果： 就拿我比较熟悉的Glean举例子，ARR最新报道过了2亿美金，我们就假设每年都是这么多收入，那现在 有1000个人，所以每个人对应收入是20万美金。这在AI行当基本会亏损。对应的事件呢，我觉得接下 来大概率会继续融资。 类似的，其他明星智能体公司也基本这样，大额融资的包括并购的... 如果就这么多，那就和上波AI有点像，还好不全是，比如： Midjourney一年赚了5亿美金的时候只有40个人，这就怎么都赚钱的。因为它流量基本还是自然流。 你比如Base44，一年350万美金，还就1个人，这怎么也是赚钱的，不太可能都是投流投出来的。 这样题目从结果看 ...

国泰海通晨报 2026 年 01 月 12 日 国泰海通证券股份有限公司 研究所 [汤蔚翔 Table_Authors] (分析师) 电话：021-38676172 邮箱：tangweixiang@gtht.com 登记编号：S0880511010007 [Table_ImportantInfo] 今日重点推荐 021-38038433 hanchaohui@gtht.com S0880523110001 张剑宇(研究助理) 021-38674711 zhangjianyu@gtht.com S0880124030031 梁中华(分析师) 021-23219820 liangzhonghua@gtht.com 宏观快报点评：《核心通胀韧性仍在》2026-01-09 12 月 CPI 同比增速+0.8%，环比+0.2%；PPI 同比增速-1.9%，环比回升至 0.2%。12 月通胀保持稳 步回升。食品价格低基数是 CPI 的主要贡献，核心通胀环比依然处于季节性上沿，同比维持稳定 （+1.2%）。12 月的金价重拾涨势是核心通胀维持韧性的关键，但中长期回升依然需要居民资产负债 表的持续修复。PPI 环比在 12 ...