一水 发自 凹非寺 量子位 | 公众号 QbitAI 10亿融资，50亿估值。 这就是李飞飞创业公司 World Labs 最新交出的答卷。 今年1月，这家炙手可热的世界模型创业新星就被传出可能正在进行一轮5亿美元新融资，但最终结果却远超预期—— 不是5亿，而是10亿美金 （约合人民币69亿元） ，而且这一轮AMD、英伟达、富达等都投了。 加上本轮融资，成立于2024年4月的World Labs可谓成长飞速—— 不到两年时间，估值从最初的10亿美金直接翻了5倍 。 就这么说吧，可能最接近World Labs融资速度的Anthropic，也花了25个月才达到50亿美元估值。 这速度，你品，你细品。 不到两年，估值一路飙升 进入2026，谁都知道世界模型很hot，但看到World Labs成长如此之快，人们或许才对这一点有了更多实感。 短短一年多时间，这家公司经历了5倍重估。 虽然这里面不乏AI教母李飞飞带来的光环效应，但不得不说，投资者对世界模型的押注决心，比预想中来得更早、也更猛。 这一点，从World Labs本轮投资阵容也可一窥。 根据官方公告，本轮融资吸引了AMD、Autodesk、Emerson Co ...

Core Viewpoint - The article discusses the ambitious plans and bold personality of Yu Hao, the founder of Chasing Technology, highlighting his extravagant company culture and aggressive market strategies aimed at achieving significant growth and innovation in the tech industry [6][8][30]. Group 1: Company Background and Leadership - Yu Hao, the founder of Chasing Technology, is known for his extravagant leadership style, including gifting gold to employees and hosting lavish company events [6][9]. - The company has set ambitious goals, including plans to expand its workforce from 20,000 to 200,000 and achieve a trillion-dollar valuation within five years [8][12]. - Chasing Technology has a strong focus on innovation, having developed a high-speed digital motor that outperforms competitors like Dyson [28][30]. Group 2: Market Position and Strategy - Chasing Technology initially gained traction by partnering with Xiaomi, which helped it penetrate the market, but this reliance limited its pricing power and profit margins [31][32]. - The company has since transitioned to developing its own brand, DREAME, and aims to establish itself in high-end markets while maintaining a focus on innovation and quality [32][33]. - By 2025, Chasing Technology is projected to achieve a valuation exceeding 20.5 billion yuan, positioning it among global unicorns [32]. Group 3: Future Aspirations and Industry Impact - Yu Hao has announced plans to diversify into new sectors, including high-end smartphones and commercial aerospace, with a notable interest in the automotive industry [33][49]. - The company aims to redefine its identity by not only competing in existing markets but also by innovating and creating new opportunities, reflecting a broader vision for technological advancement [36][56]. - Yu Hao's approach emphasizes the importance of cross-industry knowledge and adaptability, aiming to build a "universal model" that can thrive across various sectors [36].

据介绍，在硬件本体层面，具身天工3.0预留多类型拓展接口，支持各类外设工具灵活接入，可快速适 配特种作业、工业制造、商用服务等多元场景需求，实现与行业主流系统的无缝衔接。 中新社北京2月10日电 (记者 吕少威 陈杭)北京人形机器人创新中心10日发布新一代通用机器人平台具身 天工3.0，该款机器人在本体稳定性、运动控制、具身大小脑协同、全自主作业等方面有了显著提升， 也是行业内首个实现触物交互式全身高动态运动控制的全尺寸人形机器人。 在软件层面，具身天工3.0基于"慧思开物"平台生态，提供完善的开发文档、工具链及低代码开发平 台，将复杂的技术能力封装为简单易用的操作流程，合作伙伴无需进行底层技术重构，即可根据自身场 景需求开展定制化二次开发，显著降低开发门槛、缩短开发周期。 在本体性能方面，具身天工3.0通过配备高扭矩一体化关节，实现了极强的肢体爆发力，可应对特种作 业中高负载需求。该款机器人还可稳定完成翻越约1米障碍、全身高爆发连续动作等复杂运动任务，同 时结合灵活的躯干与上肢运动范围，机器人可支持跪地作业、弯腰转身操作等复杂灵巧操控场景。 得益于多自由度协同的肢体设计，具身天工3.0实现了各关节的精准联动 ...

Core Viewpoint - The article emphasizes the rapid growth and potential of the embodied intelligence sector, highlighting significant investments and innovations, particularly from the company Zhujidongli, which recently secured $200 million in funding [11][12]. Group 1: Industry Overview - In 2025, the embodied intelligence sector saw 168 companies receive funding totaling 32.9 billion, a year-on-year increase of 291% [3]. - The Ministry of Industry and Information Technology has identified humanoid robots as a key driver for the development of the embodied intelligence industry [3]. - Despite the optimism, concerns about a potential "winter" for the industry are emerging due to high valuations and challenges in technology and application [3][39]. Group 2: Company Innovations - Zhujidongli is recognized for its unique approach in the embodied intelligence space, showcasing its humanoid robot Oli during a government visit [5]. - The company introduced the "Embodied Intelligence System LimX COSA," which acts as an intelligent operating system for humanoid robots, enabling them to think and act autonomously [6][7]. - The TRON 2 robot, capable of switching between multiple forms, is designed to reduce the cost of exploring various application scenarios, making it a versatile tool for businesses [8][11]. Group 3: Market Position and Strategy - Zhujidongli's focus is on original innovation, avoiding trends and competition based solely on parameters, aiming instead for practical applications and user value [11][42]. - The company is strategically positioned to leverage its technological advancements in the humanoid robot sector, emphasizing the importance of product experience over mere technical demonstrations [17][32]. - The founder believes that 2026 will mark a pivotal year for the industry, with a shift towards practical applications and proof of concept validations [39][49]. Group 4: Future Outlook - The company anticipates that 2026 will be a year of significant breakthroughs in embodied intelligence, particularly in video data training and reinforcement learning [49][50]. - The focus will be on achieving product-market fit and expanding the product matrix, with an emphasis on the Agentic OS as a key technological advancement [55][56]. - The founder asserts that true competition in the sector will revolve around creating meaningful products and solutions rather than merely competing on technology [57][58].

Core Insights - Ant Group's LingBot-VA represents a significant advancement in general robotics, enabling robots to predict future actions before executing them, thus moving beyond traditional "observe-react" models [1][3][5] Group 1: Technological Advancements - LingBot-VA introduces a causal video-action world model that allows robots to imagine future scenarios before taking action, enhancing decision-making capabilities [3][5] - The model retains memory during long sequence tasks, demonstrating strong state awareness and the ability to adapt to new tasks with minimal examples [5][6] - The architecture separates visual understanding, physical reasoning, and action control, improving sample efficiency and generalization [6][7] Group 2: Performance and Testing - In real-world tests, LingBot-VA successfully handled complex tasks such as preparing breakfast and high-precision actions like cleaning test tubes, showcasing its stability and adaptability [21][22][25] - The model achieved a success rate of 92.93% in easy scenarios and 91.55% in hard scenarios on the RoboTwin 2.0 benchmark, outperforming competitors [28][30] - In the LIBERO benchmark, LingBot-VA achieved an average success rate of 98.5%, setting a new state-of-the-art record [30][31] Group 3: Industry Impact - The continuous open-sourcing of LingBot-VA and its predecessors signals a strategic move by Ant Group to establish leadership in the global robotics field [34][38] - The integration of video as a medium for reasoning and action in robotics represents a paradigm shift, addressing challenges in long tasks and complex environments [35][36] - The emergence of LingBot-VA positions world models as a central capability in robotics, evolving from mere action to thoughtful action [36][40]

Core Viewpoint - The article discusses the groundbreaking advancements made by Ant Group's LingBot-VA, which represents a significant leap in robot control by enabling robots to predict future actions before executing them, thus enhancing their decision-making capabilities [2][11][56]. Group 1: Technological Innovations - LingBot-VA introduces a causal video-action world model that allows robots to visualize future scenarios before taking action, moving beyond the traditional "observe-react" model [6][12]. - The model features strong memory retention, enabling it to remember previous actions during long sequences, and demonstrates high adaptability with minimal training samples [8][10]. - The architecture separates visual understanding and action control, enhancing sample efficiency and generalization capabilities [14][15]. Group 2: Performance and Testing - In real-world tests, LingBot-VA successfully handled complex tasks such as preparing breakfast and manipulating delicate objects, showcasing its stability and precision [34][36]. - The model achieved a success rate of 92.93% in the RoboTwin 2.0 benchmark for easy tasks, outperforming competitors by 4.2% [40]. - In the LIBERO benchmark, LingBot-VA set a new state-of-the-art record with a 98.5% average success rate [42]. Group 3: Industry Impact - The continuous open-sourcing of LingBot-VA and its related projects signals a shift towards a video-centric approach in robotics, where video becomes a medium for reasoning and action [46][48]. - The advancements in LingBot-VA position world models as a central capability in robotics, evolving from mere action to thoughtful decision-making [49][56]. - The ripple effect of these innovations is evident, with increased attention from global tech companies and media, indicating a strategic move in the competitive landscape of robotics [52][56].

Core Viewpoint - The article discusses the rapid rise of embodied intelligence and the skepticism among industry practitioners regarding the feasibility of achieving true general-purpose robots, emphasizing the need for a realistic assessment of the technology's current state and future potential [2][3]. Group 1: Industry Insights - The conversation highlights the significant financial backing from national strategies that have propelled the development of brain-like robots, indicating a long-term commitment to this technology [5]. - Entrepreneurs express a shared vision of enhancing the intelligence and functionality of various smart devices and robots, focusing on practical applications rather than strictly adhering to the concept of "embodiment" [5]. - The discussion reveals a consensus that the term "general-purpose" in robotics is overly broad and often unrealistic, suggesting a more focused approach on specific operational domains (ODD) for practical advancements [6][7]. Group 2: Technology and Commercialization - The panelists caution against conflating impressive demos with actual commercialization, stressing that demos do not equate to market-ready products and that a clear understanding of the path from demo to commercial viability is essential [12][13]. - There is a recognition of the evolving nature of demos, which are now often aimed at ecosystem partners rather than end-users, leading to potential misunderstandings about their commercial applicability [14]. - The importance of balancing algorithm development with engineering capabilities is emphasized, with a call for teams to possess a diverse skill set that includes both technical and operational expertise [15][16]. Group 3: Future Outlook - Looking ahead to 2026, the key to survival for companies in the embodied intelligence sector will be their ability to transition from pure technology development to effective business operations and commercialization strategies [19]. - The discussion identifies three critical cycles in the industry: hardware, scenario, and data cycles, with a focus on the scenario cycle as particularly important for the upcoming year [20]. - The competitive landscape is expected to shift in 2026, with companies beginning to vie for orders and market share, indicating a more aggressive commercial environment [20].

Core Viewpoint - The article discusses the establishment of a new robotics company, Sharpa, by the co-founders of Hesai Technology, highlighting the significant potential and challenges in the robotics industry compared to the electric vehicle sector [4][10]. Company Overview - Sharpa was founded by Hesai's co-founders, including CTO Xiang Shaoqing and CEO Li Yifan, focusing on developing general-purpose robots and core components [4][6]. - The company operates independently from Hesai, with a team of over 100 members primarily recruited externally [4]. Market Potential - The general robotics market is estimated to have a trillion-dollar potential, with the founders aiming to create a world-class AI company [7]. - The founders believe that the AI era presents opportunities similar to those seen in the internet and new energy sectors, potentially leading to the creation of companies with valuations in the hundreds of billions [7]. Product Development - Sharpa's first product, the SharpaWave dexterous hand, is set to be mass-produced by October 2025, featuring 22 degrees of freedom and priced at several tens of thousands of dollars [8]. - The second product will be a general humanoid robot named Sharpa North, with a focus on versatility and operational capabilities [8][9]. Industry Challenges - The founders acknowledge that while the opportunity in robotics is ten times greater than in new energy vehicles, the difficulty of development is a hundred times higher [10]. - The robotics industry is still in its exploratory phase, with over 300 startups working on various applications and technologies [9]. Strategic Vision - The founders intend for the AI company to enhance Hesai's market reach and capabilities, potentially increasing its market ceiling by tenfold [8]. - They emphasize the importance of creating robots that can handle complex tasks and adapt to changing environments, bridging the gap between human capabilities and automation [9].

Group 1 - ByteDance announced the 2.0 brand upgrade for "Kouzi," introducing new features such as Agent Skills and Agent Plan [1] - Hesai Technology's founding team has entered the general robotics sector [1]

Group 1 - Hesai Technology and Sharpa operate as independent companies without any equity or business control relationship [1] - The AI company Sharpa is expected to broaden the team's perspective and provide feedback to Hesai [1] - The future market for robotic LiDAR may surpass that of automotive applications, and Hesai plans to develop other components needed for robotics beyond LiDAR [1] Group 2 - The founders of Hesai Technology, including Xie Shaoqing, Li Yifan, and Sun Kai, have jointly established Sharpa [2] - Sharpa focuses on developing general-purpose robots and their core components [2]