编辑部 整理自 MEET2026 量子位 | 公众号 QbitAI 从大模型智能的"语言世界"迈向具身智能的"物理世界"，仿真正在成为连接落地的底层基础设施。 在本次量子位MEET2026智能未来大会上，光轮智能联合创始人兼总裁 杨海波 给出了他的观察： 具身智能的规模远大于文本与视觉模型，因为数据维度更真实、更复杂。 这也就意味着，具身智能时代的核心，不是算法本身，而是它所依赖的数据是否有效、可扩展——仿真是唯一能够解决数据问题的方案。 在仿真策略的路上，会遇到仿真不真实、Sim2Real不可靠等行业痛点， 光轮智能正在通过自研的一整套"测量、生成、求解"仿真基础设施来 解决这些问题 ，为具身智能提供数据、训练、评测的全流程解决方案。 △ 杨海波指出光轮智能深耕合成数据领域 另外杨海波还进一步指出， 仿真不是孤立的技术工具，需要以真实产业需求为锚点，通过应用场景构建生态。 其中， 具身仿真资产制作是生态的源头活水 ，依托自动化物理测量与生成技术，产出高物理真实的规范化数据资产，为具身训练提供核心燃 料； 大规模RL训练则通过并行的虚拟场景让智能体高效试错学习，将数据价值转化为具身实际技能 ，同时反向打磨仿真 ...

Core Insights - The article highlights the recent funding news for Lightwheel Intelligence, a company specializing in simulation and synthetic data, which has completed several hundred million yuan in Series A and A+ financing [2] - The funding will primarily be used for scaling delivery capabilities, investing in technology research and development, and attracting high-level talent [2] - Lightwheel has established partnerships with leading companies in the industry, including NVIDIA, Google, and Toyota, and has seen exponential growth in order demand, with annual revenue surpassing 100 million yuan [2] Group 1: Industry Context - The article discusses the significance of Physical AI as a multi-billion dollar business addressing a multi-trillion dollar opportunity, as highlighted by NVIDIA's recent financial report [3][4] - NVIDIA's CEO emphasized that Physical AI represents the next growth engine for the company, indicating a strong market potential [4] Group 2: Challenges in Physical AI - A major challenge facing Physical AI is the data scarcity for developing robotic foundational models, which differs significantly from large language models that have ample internet text data for pre-training [9] - The lack of large datasets for physical world interactions poses a bottleneck for both embodied intelligence and world model development [9][10] Group 3: Solutions Offered by Lightwheel - Lightwheel aims to address the data shortage through simulation, allowing robots to learn faster in a simulated environment compared to real-world learning [12] - The company provides a comprehensive platform for robotics users to generate high-quality synthetic data and conduct simulations, effectively creating a "playground for robotics users" [13][15] - Lightwheel's technology integrates with NVIDIA's platforms, offering a rich library of physically accurate assets for various applications, ensuring that robots can transfer learned skills to real-world scenarios [16][19] Group 4: Strategic Partnerships - The frequent interactions between Lightwheel and NVIDIA underscore their strategic partnership, with Lightwheel contributing to NVIDIA's ecosystem by providing synthetic data support for various models [20] - This collaboration not only enhances Lightwheel's technological credibility but also positions it within the top-tier robotics ecosystem globally [20] Group 5: Future Outlook - Lightwheel's CEO expressed optimism about accelerating the development of the $50 trillion robotics industry through simulation technology [21] - The company plans to focus on building scalable delivery capabilities to meet the rapidly growing market demand, positioning itself as a leading data infrastructure provider in the Physical AI and world model data market [23]

衡宇 发自 凹非寺 量子位 | 公众号 QbitAI 刚刚，一家AI公司的融资引发了圈内热议。 Why？因为它与具身智能息息相关，也与通往物理AI的世界模型密不可分。更准确来说，完成融资的这家公司是站在二者相关生态上的关键供 应链公司——仿真合成数据公司。 量子位最新获悉， 仿真合成数据公司光轮智能，刚刚完成数亿元A轮、A+轮融资 。 此次披露的投资方里，既有东方富海、九派资本等机构投资者，也有三七互娱、琥珀资本等产业方。老股东辰韬资本也持续加注。 而同样受关注的是它合作的客户，既有英伟达、谷歌、阿里、字节，也有Figure AI、1X Technology、智元机器人、银河通用，还有 Toyota，BOSCH、比亚迪、吉利…… 一己之力，串起了整个AI生态 。 有消息称，这家全球唯一专注仿真合成数据的技术公司， 营收已突破亿元大关 。 而作为全球首家把生成式AI融入仿真技术的公司， 光轮智能的创始人是圈内声名卓著的大佬谢晨 ——之前英伟达、Cruise及蔚来的仿真负责 人。 最近一次出圈，则因为与黄仁勋女儿Madison Huang的首秀对谈，谈论的话题还是风口上的物理AI…… 物理AI是黄仁勋在2025年 ...