机器人行业的价值重构正在加速进行。机器人的定价权正在从硬件本体向大脑模型转移，资本和政策正加速向掌握模型能力的企业集中[1]。2026年上半年国内具身智能赛道融资总额达935亿元，较2025年上半年增长超过5倍[1]。这一融资热潮反映出行业对模型驱动力量的共识，DeepSeek R1的发布证明了模型能力而非算力才是估值溢价的核心，这一逻辑正在具身智能领域加速复现[1]。
全球头部企业和初创公司的融资动向印证了这一趋势。Figure AI投后估值390亿美元，累计融资19亿美元，投资方包括微软、英伟达、OpenAI和贝索斯[1]。特斯拉Optimus确认即将进入量产，V3版本年产100万台、V4版本年产1000万台[1]。国内市场方面，一家学术背景的具身大脑团队从天使轮估值数亿元到跻身10亿美金独角兽仅需8到10个月[1]，国内已有8家具身智能企业跻身200亿元估值俱乐部[1]。2026年7月，港交所优化18C章上市制度，降低科技企业上市门槛[1]，为这些企业提供了新的融资通道。
数据驱动成为行业竞争的新前沿。全球最大规模的具身智能数据采集仅百万条量级，而头部车厂单天回流驾驶数据约1亿条[1]，反映出现有数据供应的巨大缺口。国家数据局将为具身智能行业提供高质量数据集作为政策发力点[1]，中国企业在数据和政策支持上构建了独特优势[1]。
The embodied artificial intelligence sector is undergoing a fundamental value restructuring, with pricing power shifting from hardware components to the artificial intelligence models that serve as robots' "brains." [1] This transition marks a critical inflection point in the industry, as capital and policy support increasingly concentrate among companies that command advanced model capabilities. [1]
In the first half of 2026, domestic embodied AI financing reached 93.5 billion yuan, surpassing the same period in 2025 by more than fivefold. [1] The sector has matured rapidly enough that eight domestic embodied intelligence companies have already reached valuations exceeding 20 billion yuan. [1] One academic-backed embodied AI team progressed from an angel round valuation of hundreds of millions of yuan to unicorn status in just eight to ten months, illustrating the accelerated growth trajectory. [1] Figure AI exemplifies the international investment momentum, with a post-investment valuation of 39 billion dollars and cumulative funding of 1.9 billion dollars backed by investors including Microsoft, Nvidia, OpenAI, and Jeff Bezos. [1]
Tesla's Optimus robot is confirmed to enter mass production, with the V3 version projecting 1 million units annually and the V4 version projecting 10 million units annually. [1] At the policy level, China's State Data Bureau has identified high-quality dataset provision as a key intervention point for the embodied AI sector. [1] This data advantage is substantial: the world's largest embodied AI data collection effort has accumulated only approximately one million samples, while leading automotive manufacturers generate around 100 million driving data points daily. [1] In July 2026, the Hong Kong Stock Exchange optimized its Chapter 18C listing rules, lowering the threshold for technology company initial public offerings. [1]