Reportedly: Beijing releases 2026-2028 embodied AI robot action plan targeting 200,000 units annual capacity and 50,000 PFlops dedicated compute
Reported by Zhidongxi · not yet confirmed by the company or a second independent outlet. We update this page when it is.
Beijing Municipal Bureau of Economy and Information Technology released the Embodied AI Robot Industry High-Quality Development Action Plan (2026-2028) on November 11, 2026. The plan targets at least 5 world-class and 10 domestic-leading humanoid robot manufacturers by 2028, with Beijing-Tianjin-Hebei region accounting for 25% of national capacity and over 60% key component localization. It also calls for over 50,000 PFlops of dedicated embodied AI compute and 200,000 units/year production capacity in Beijing, Zhidongxi reported.
China context
- Original name
- 北京市经济和信息化局
- Outside China
- Not stated in the sources yet
- Claims
- Company-reported; not yet independently evaluated
- For builders
- Developers of embodied AI models and robot systems may find new opportunities in Beijing's push for data collection, simulation, and deployment infrastructure, but must navigate local regulations and potential localization requirements.
- For investors
- Investors should monitor which companies receive support under the plan, as it may signal government backing and reduce risk for selected humanoid robot and AI infrastructure ventures in the Beijing-Tianjin-Hebei region.
Translated from Chinese. Quotes and facts link to the original sources.
The Beijing Municipal Bureau of Economy and Information Technology published the Embodied AI Robot Industry High-Quality Development Action Plan (2026-2028) on November 11, 2026. The plan sets goals for 2028: cultivate at least 5 world-class and 10 domestic-leading humanoid robot manufacturers; achieve world-class embodied AI models, robot bodies, and key technologies; have the Beijing-Tianjin-Hebei region account for 25% of national robot production capacity and over 60% key component localization; and establish a comprehensive safety governance system. It outlines six major projects: embodied AI leap, Beijing-Tianjin-Hebei industrial chain collaboration, technology innovation source, large-scale application promotion, industrial ecosystem leadership, and safety governance first. Specific targets include adding over 50,000 PFlops of dedicated embodied AI computing clusters, collecting over 1 million hours of high-quality data from at least 100 real scenarios, and forming a production capacity of 200,000 units per year in Beijing. The plan also includes special support policies, increased funding for key technology research, innovative product development, pilot testing, and typical scenario applications, as well as talent incentives such as Beijing hukou, children's education, housing, and medical care for urgently needed embodied AI robot professionals.
The plan emphasizes multi-technology route development including Vision-Language-Action (VLA) models, world models, and reinforcement learning. It targets breakthroughs in multi-source information fusion, physical space understanding, and human emotion recognition to advance from perception to cognition. It also aims to improve real-time reasoning, cross-embodiment transfer, self-learning evolution, group collaboration, long-term memory, lightweight local deployment, and model interpretability. Data collection innovations include no-embodiment collection and first-person view collection, high-fidelity simulation data generation, autonomous exploration data collection, human motion mapping, and multimodal unified representation.
Humanoid robot manufacturers in Beijing will face new policy-driven targets for production capacity and component localization, potentially accelerating investment and consolidation in the Beijing-Tianjin-Hebei region. The plan's 50,000 PFlops dedicated compute cluster and 1 million hours of data collection requirements will increase demand for specialized AI infrastructure and data services, benefiting suppliers in those areas.
The plan creates opportunities for companies providing embodied AI compute infrastructure, data collection and annotation services, and humanoid robot components. It also signals potential government subsidies and talent incentives for qualifying enterprises in Beijing, which could lower operational costs and attract skilled workers.
The next verifiable signal is the release of specific implementation rules or funding guidelines by Beijing municipal departments, which would clarify how the 200,000 units/year capacity and 50,000 PFlops compute targets will be achieved. Another signal is the announcement of the first batch of supported enterprises or projects under the plan.