New national-level policies are providing a systematic framework to leverage data for advancing artificial intelligence, directly addressing a critical bottleneck in the embodied intelligence sector by accelerating domestic dataset development.

With the humanoid robot industry entering its first year of mass production in 2026, these timely measures offer essential infrastructure support for the industry's next phase of growth. The key points from the report are outlined below.

Policy Overview and Strategic Focus

On June 3, 2026, the National Data Administration officially issued the "Implementation Plan for Promoting the Construction of High-Quality Industry Datasets." The plan clearly aims to establish a batch of high-quality, application-verified datasets covering key sectors by the end of 2028, creating exemplary application scenarios for data-driven AI innovation.

The initiative deploys six major special actions: strengthening foundations and expanding capacity, tackling data annotation challenges, improving quality and efficiency, enabling applications, enhancing management services, and unlocking value. This marks the first systematic national-level deployment focused on empowering AI development through data.

Significantly, the plan includes embodied intelligence as a key supported area for innovation, creating direct synergy with the development of the robotics industry.

Addressing Core Industry Challenges

Starting from a top-level design perspective, the plan strengthens the role of datasets to tackle the core pain point of data scarcity in the embodied intelligence field. Unlike traditional large language models that can utilize vast amounts of internet text, training embodied intelligence models heavily relies on multimodal physical data encompassing vision, force sensing, motion trajectories, and physical interaction outcomes.

The previous severe shortage of such embodied data, often described as a "data famine," has been a major bottleneck limiting breakthroughs in the generalization capabilities of embodied intelligence.

The year 2026 marks the official start of mass production for the humanoid robot industry. With overseas projects like Tesla's Optimus V3 expected to begin production in the second half of the year, and domestic companies like Ubtech and Unitree accelerating their application scenarios, the industry's competitive focus is shifting from hardware capabilities to the "embodied brain" and data reserves. The policy's introduction is well-timed, providing crucial infrastructure support for the industry's advancement.

Momentum in Dataset Construction and Capital Inflow

Prior to this, the National Data Administration had already designated seven pilot cities for data annotation: Chengdu, Shenyang, Hefei, Changsha, Haikou, Baoding, and Datong. By the first half of 2025, these seven data annotation bases had constructed 524 datasets, served 163 large models, and generated over 8.3 billion yuan in related output value for the data annotation industry.

Since May 2026, there has been a concentrated surge in Series A funding within the embodied data sector. Jianzhi Robotics completed a financing round worth hundreds of millions of yuan, becoming the highest-funded company in the "body-less data" segment. Companies like Luming Robotics, Guanglun Intelligence, and Heshuju have also successively secured substantial funding.

The capital rationale is clear: while most companies in the embodied intelligence field are not yet profitable, all players must procure data, making data service providers the essential "shovel sellers" in the industry chain.

Investment Recommendations

The report suggests focusing on the robotics industry chain, with attention on the following areas: 1) Core components and complete bodies for humanoid robots: recommendations include 恒立液压 (Hengli Hydraulic), 长盈精密 (Changying Precision), 双环传动 (Shuanghuan Transmission), 优必选 (Ubtech), 安培龙 (Ampelion), 杭叉集团 (Hangcha Group), and 绿的谐波 (Leader Harmonious Drive). Suggested companies to watch include 浙江荣泰 (Zhejiang Rongtai), 峰岹科技 (Fengtiao Technology), 科达利 (Kedali), and 奥比中光 (Orbbec). 2) Industrial robots: 埃斯顿 (Estun) is recommended. 3) Collaborative robots: suggested watchlist includes 华沿机器人 (Huayan Robot). 4) Autonomous Mobile Robots (AMR): suggested watchlist includes 凯乐士科技 (Galaxy Technology) and 极智嘉 (Geek+). 5) Model, computing power, and data segments: suggested watchlist includes 卧安机器人 (Woan Robot) and 地平线机器人 (Horizon Robotics).

Key Risks to Consider

Potential risks include slower-than-expected technological progress, delays in industrial commercialization, and increased market competition.