中金研究 随着Genie3等AI生成物理世界模型的亮相，市场掀起了一场关于"游戏行业将被颠覆"的热议。我们本篇报告的核心观点是：市场高估了AI模型对游戏 产业的短期颠覆力，同时低估了其作为工具的长期赋能价值。Genie3所代表的世界模型，本质上是高效率的内容生成工具，但它未触及游戏的核心 ——玩法设计、数值平衡与长线运营。头部游戏厂商仍具备较强内容护城河。 点击小程序查看报告原文 Abstract 摘要 世界模型是游 戏行业的 颠覆性生产力吗？在游戏行业应用的发展方向如何？ 我们认为，当前世界模型在游戏制作中应用仍处于早期，起到局部赋能作 用，在预研、3D资产制作上提效较为明显，未来更可能作为单场景生成工具而非综合化游戏基座。 AI新技术会否重塑游戏产业链？谁将受益？ 我们判断，最直接面临冲击的是产业链中标准化程度较高的外包服务环节，主要包括美术资产制作、基础音 频录制、部分代码编写等。利润池的迁移方向之一是游戏产业链上游的创意与管理环节。我们认为： ► 头部游戏厂商 仍具备较强内容护城河。中长期来看善于利用AI赋能创作、挖掘AI潜能并持续巩固IP与玩法优势的公司有望在当前变革中拥有持续性优 势。 风险 游戏 ...

March 22, 2026 09:00 PM GMT Embodied AI | North America AI's Next Challenge: Modeling the Physical World. LLMs train on, process, and generate text in its many forms. While they have proven to be powerful tools for white-collar tasks such as coding, search, and writing, AI's broader potential may lie in the physical world, which is governed by the laws of physics - substances, thermodynamics, fluid dynamics and the behavior of light - in a constantly changing 3-dimensional space. 'World Models' are AI syste ...

Group 1 - The article highlights the significant investment in AI startups, particularly focusing on Yann LeCun's AMI Labs, which raised $1.03 billion (approximately 70.87 billion RMB) from notable investors including KKR and Bezos Expeditions [2] - AMI Labs aims to develop "World Models" rather than large language models (LLMs), focusing on creating systems that can learn abstract representations from real-world sensor data [7][9] - The company has a strong founding team, including former Meta AI researchers and professors from prestigious universities, indicating a robust technical foundation [5][7] Group 2 - The rise of "World Models" is seen as a new wave of technological disruption in AI, with other prominent figures like Fei-Fei Li also entering the space, as evidenced by the $1 billion funding for her company, World Labs [3][10] - The article discusses the increasing trend of tech talent leaving large companies to start their own ventures, which is driving innovation in the AI sector [14][15] - The potential market for "physical AI" is projected to reach $90 trillion, indicating a massive opportunity for companies focusing on this technology [12]

Core Viewpoint - The article discusses the concept of "world models" in AI, emphasizing their potential to enable machines to understand, predict, and interact with the world, moving towards achieving Artificial General Intelligence (AGI) [4][6]. What is a World Model? - The definition of a world model is still evolving, but it is rooted in the idea that humans use mental models to predict outcomes based on their understanding of the world [7][8]. - World models are essential for AI to achieve true intelligence, allowing machines to simulate and predict the consequences of their actions [10][12]. - The concept has been explored since the 1940s, with significant developments in AI and reinforcement learning leading to the formalization of world models in recent years [9][17]. - A world model consists of three core components: observation of the world, prediction of future states, and learning to act within an internal representation of the world [18][24]. Why Study World Models? - World models differ from large language models (LLMs) in their objectives, training data, and outputs, focusing on dynamic understanding and interaction with the environment [28][30]. - The limitations of LLMs have prompted a renewed interest in world models, as they are seen as a necessary step towards achieving AGI [32][40]. - The emergence of multi-modal technologies has made it feasible to train effective world models, which require vast amounts of visual and action data [44][46]. Current Approaches to World Models - The industry is exploring various approaches to world models, which can be categorized into three layers: foundational theories, representation forms, and training objectives [49][50]. - The focus on world generation is crucial, as it lays the groundwork for understanding how the world evolves over time and how AI can interact with it [54][56]. - Two main technical routes for world generation are video generation and 3D spatial generation, each with its own advantages and challenges [56][70]. Impact on Key Industries - The robotics industry stands to benefit significantly from world models, as they can enable robots to understand and predict their environment, enhancing their adaptability and functionality [106][109]. - In autonomous driving, world models can improve the ability of systems to predict future scenarios, addressing current limitations in perception and decision-making [110][113]. - Wearable devices can evolve from simple data recorders to intelligent companions that understand and interact with the user's environment, fundamentally changing human-device relationships [114][116].

Core Viewpoint - The article discusses the importance of 3D as a structured interface for AI systems to interact with the physical world, emphasizing its role in enhancing communication between humans and machines, as well as among machines themselves [2][24]. Group 1: The Role of 3D in AI and Space - 3D serves as a structured representation that allows for the generation, modification, and management of spatial environments, akin to how code functions in programming [2][3]. - The evolution of hardware, particularly GPUs, has enabled advanced techniques like NeRF and Gaussian splatting, which facilitate the creation and rendering of large-scale 3D representations [10][11]. - The integration of 3D representations into existing workflows can enhance the fidelity of environments and support new applications such as digital twins [11][23]. Group 2: Interaction Between Humans and Machines - 3D provides a natural medium for human interaction, aligning with how humans navigate and understand spatial relationships [19][22]. - The separation of 3D representation from rendering allows for efficient updates and modifications, similar to how code can be edited without rewriting every output [20][21]. - The structured nature of 3D outputs enables them to be checked, validated, and reused, fostering collaborative workflows between designers and AI systems [18][23]. Group 3: Future Directions and Applications - The article highlights a paradigm shift towards programmable, data-driven spatial systems that can support realistic environments and new application categories [23]. - The development of tools like Marble and Chisel aims to enhance the creation and manipulation of 3D worlds, allowing for detailed visual effects and structural control [25]. - The ongoing evolution in this field suggests a growing reliance on 3D as a universal interface for communication and collaboration among humans, agents, and software [24][23].

Investment Rating - The report maintains a "Buy" rating for the companies mentioned, including Weike Technology (301196.SZ), Nengke Technology (603859.SH), Hehe Information (688615.SH), and Maixinlin (688685.SH) [8][53]. Core Insights - The report highlights significant advancements in AI technology, particularly with the release of LLaDA2.1 and Gemini3.1Pro, which set new standards in multimodal AI applications [1][35]. - LLaDA2.1 achieves a peak speed of 892 Tokens/second in complex programming tests, showcasing a balance between speed and quality through innovative technical breakthroughs [3][23]. - Gemini3.1Pro demonstrates exceptional reasoning capabilities, scoring 77.1% in the ARC-AGI-2 test, significantly outperforming its predecessor and competitors [35][36]. Summary by Sections Computing Power Dynamics - The report notes stable rental prices for computing power, with specific configurations priced at 28.64 CNY/hour for Tencent Cloud and 31.58 CNY/hour for Alibaba Cloud [22]. - LLaDA2.1, released in February 2026, includes versions with 16 billion and 100 billion parameters, breaking the speed barrier for diffusion language models [23][26]. - The model introduces a novel error-correcting editing mechanism, enhancing generation efficiency without sacrificing quality [27][29]. AI Application Dynamics - Gemini's weekly traffic increased by 4.31%, indicating growing user engagement [33]. - Gemini3.1Pro supports up to 1 million tokens in context, achieving high scores in long-text processing and significantly reducing hallucination rates compared to previous models [40][41]. AI Financing Trends - WorldLabs completed a $1 billion funding round, with notable investors including AMD and NVIDIA, aimed at advancing spatial intelligence technologies [46][48]. - The funding will accelerate the development of their flagship product, Marble, which focuses on generating high-fidelity 3D worlds for various applications [47][49]. Investment Recommendations - The report suggests focusing on companies that are expanding their computing power capabilities, such as Maixinlin and Weike Technology, as well as those excelling in AI applications like Hehe Information and Nengke Technology [52].

Investment Rating - The report maintains a "Buy" rating for the companies mentioned, including Weike Technology (301196.SZ), Nengke Technology (603859.SH), Hehe Information (688615.SH), and Maixinlin (688685.SH) [8][53]. Core Insights - The LLaDA2.1 model has achieved a technological breakthrough, featuring two versions with 16 billion and 100 billion parameters, demonstrating a peak speed of 892 Tokens/second in complex programming tests [3][23]. - Gemini3.1Pro, released by Google DeepMind, has set a new standard in multimodal AI, achieving a score of 77.1% in the ARC-AGI-2 test, more than doubling the performance of its predecessor [3][35]. - WorldLabs has completed a new funding round of $1 billion, with investments from major firms like AMD and NVIDIA, focusing on spatial intelligence and large world models [4][46]. Summary by Sections Computing Power Dynamics - The rental prices for computing power remain stable, with specific configurations priced at 28.64 CNY/hour for Tencent Cloud and 31.58 CNY/hour for Alibaba Cloud [22]. - The LLaDA2.1 model's release marks a significant advancement in diffusion language models, providing a new feasible path for the development of large language models [23][32]. AI Application Dynamics - Gemini's weekly traffic increased by 4.31%, with notable performance in user engagement metrics [33][34]. - Gemini3.1Pro excels in reasoning capabilities and has shown significant improvements in long-context processing, handling up to 1 million tokens [40][41]. AI Financing Trends - WorldLabs' recent funding will accelerate its research and development in spatial intelligence, with a focus on applications in robotics and AR/VR [46][48]. Investment Recommendations - The report suggests a focus on companies that are expanding their computing power capabilities, such as Maixinlin (688685.SH) and Weike Technology (301196.SZ), as well as those in AI-driven sectors like Hehe Information (688615.SH) and Nengke Technology (603859.SH) [52].

Core Viewpoint - The article highlights the significant rise in the stock prices and market valuations of major AI model companies, particularly Zhiyu and MiniMax, driven by advancements in their AI models and new financing rounds in the industry [1][5][10]. Group 1: Stock Performance - Zhiyu, known as the "first global AI model stock," saw its stock price increase by over 42% on the first trading day of the Year of the Horse, with a market capitalization exceeding 320 billion HKD [1]. - Since February 9, Zhiyu's stock has surged over 250% over seven trading days, while MiniMax's stock has risen over 110% in the same period [1][5]. - As of February 20, Zhiyu's cumulative increase in February is over 220%, and MiniMax's is over 105% [5]. Group 2: Model Developments - Zhiyu recently launched its flagship model GLM-5, designed for complex system engineering and long-range agent tasks, achieving performance levels comparable to top international models [6]. - GLM-5 ranks fourth globally and first among open-source models, marking a significant achievement for domestic models [6]. - MiniMax introduced its latest flagship programming model, MiniMax M2.5, which is designed for agent scenarios and competes with leading international models [8]. Group 3: Financing and Valuation - OpenAI is expected to raise over 100 billion USD in its latest financing round, potentially increasing its valuation to over 850 billion USD, a 70% increase from the end of last year [3][10]. - The domestic AI model company Kimi has seen its valuation double in its latest financing round, indicating strong investor interest [3][10]. - World Labs, founded by AI pioneer Fei-Fei Li, completed a new financing round of 1 billion USD, focusing on enhancing spatial intelligence through world models [11]. Group 4: Market Trends and Future Outlook - The article notes a competitive landscape in the global AI model sector, with major companies investing heavily to capture consumer markets [10]. - The integration of AI with robotics is expected to drive significant growth in the coming years, with predictions of a substantial increase in market demand for embodied intelligence [9]. - The Chinese AI market is projected to grow from 93.7 billion CNY in 2022 to 160.7 billion CNY in 2024, with a compound annual growth rate of 31% [11].

Core Insights - World Labs, founded by Fei-Fei Li, has completed a new financing round of $1 billion, bringing its valuation to $5 billion within just 16 months of establishment [1][2][4]. - The company has attracted significant investments from major players like NVIDIA and AMD, as well as new investors including Autodesk, which invested $200 million [3][4]. - World Labs aims to focus on robotics and scientific discovery, actively recruiting engineers to develop core technologies for autonomous navigation and 3D reconstruction [4][5]. Company Overview - World Labs was founded by Fei-Fei Li, a prominent figure in AI, who previously initiated the ImageNet project that significantly advanced deep learning in computer vision [5][6]. - The company has raised a total of approximately $2.3 billion in funding since its inception, with its latest round pushing its valuation to $5 billion [4][6]. - The founding team includes notable scientists and alumni from prestigious institutions, reflecting a strong academic and professional background [6][8]. Product and Technology - World Labs launched its first commercial product, Marble, which generates persistent 3D worlds from various input formats using advanced technologies like Neural Radiance Fields (NeRF) [4][5]. - The business model is based on a freemium approach, allowing users to generate content for free up to four times, with paid plans available for $95 per month [4]. - The company is set to collaborate with Autodesk to enhance its research and modeling capabilities, aiming to build physical AI that enhances human creativity [5][8]. Industry Context - The AI sector is experiencing a wealth creation surge, with companies like Anthropic and Databricks achieving valuations of $380 billion and over $134 billion, respectively [9][11]. - In China, AI companies like Zhiyuan and MiniMax have seen their market capitalizations soar, reflecting the global trend of rising valuations in the AI industry [11]. - The current investment landscape prioritizes "AI content" as a key evaluation criterion for projects, indicating a strong market focus on AI-driven innovations [11].

来源：环球网 作为人工智能领域的重要探索方向，"空间智能"由李飞飞率先提出并进行系统性研究，她将人类智能归 结为语言智能与空间智能两大核心，其中空间智能是生物理解、推理周边世界并实现感知与行动关联的 基础能力，经过数百万年演化，成为人类推理、行动、发明创造的重要支撑，从日常的生活互动到城市 的可视化设计，均离不开空间智能的作用。在李飞飞看来，当前人工智能在语言智能领域已取得显著发 展，但要实现技术的进一步突破，就需要让AI突破2D屏幕的限制，拥有理解、生成并互动3D世界的能 力，实现从"能看会说"到"能行动"的跨越。英伟达相关人士也评价，空间智能已成为计算机视觉和实体 智能体发展的下一个前沿。 自成立以来，World Labs便以打造AI领域的革命性"世界模型"为目标，致力于让AI在3D世界中实现感 知、生成与互动，计划颠覆现有叙事方式、创意产出、机器人技术及科学发现的固有边界。2025年底， 企业推出了首款核心产品Marble，这款工具可通过图像、视频甚至纯文本等简单输入，快速生成具备空 间连贯性、高保真度且可持久存在的3D虚拟世界，成为"空间智能"技术落地的重要成果，也让文字、 图像生成3D虚拟宇宙从概念 ...