Chinese humanoid robot companies are pulling in hundreds of millions of dollars from investors, some of it traceable to American capital, at the same time Washington is tightening restrictions meant to limit Beijing’s access to advanced technology. The collision between surging private investment and escalating export controls is shaping a contest where China already dominates global shipments and U.S. policy may be inadvertently steering Chinese firms toward the software and AI layers that matter most for next-generation robotics.
Massive Funding Rounds Fuel Chinese Robotics
Two recent fundraising events illustrate the scale of capital flowing into Chinese humanoid robot development. Leju Robotics raised more than $200 million ahead of a planned initial public offering, according to a company statement. Separately, X Square Robot closed a Series A++ round of about $140 million, equivalent to RMB 1 billion, with headline investors including ByteDance and HongShan. X Square Robot said the capital will accelerate work on a foundation model it calls the “robot brain,” a general-purpose AI system designed to give humanoid machines the ability to learn and adapt across tasks.
These are not small seed checks. Combined, the two rounds exceed $340 million and signal that private investors view humanoid robotics as a near-term commercial opportunity rather than a distant research bet. The IPO track for Leju Robotics suggests its backers expect public-market appetite for the sector, while ByteDance’s participation in X Square Robot ties one of China’s largest technology conglomerates directly to the race for physical AI. The size and speed of these rounds matter because they give Chinese startups the runway to iterate on hardware and software simultaneously, compressing development timelines that Western competitors are still stretching across multiple funding cycles.
China Ships 90 Percent of Global Humanoid Robots
The funding surge arrives on top of an already commanding production lead. China accounted for 90% of the roughly 13,000 humanoid robots shipped globally last year, far ahead of U.S. rivals including Tesla. That ratio reflects years of state-backed industrial policy, cheap manufacturing infrastructure, and a domestic market eager to deploy robots in factories, logistics hubs, and public-facing service roles. For American companies still prototyping or running limited pilot programs, the gap is not just about unit volume but about the feedback loop that real-world deployment creates: every robot in a Chinese warehouse generates data that trains the next generation of control software.
The 90% figure also reframes the policy debate in Washington. Export controls on advanced semiconductors were designed to slow China’s progress on military-grade computing, but the humanoid robot sector depends as much on software intelligence and sensor integration as it does on cutting-edge chips. Chinese firms that cannot easily source the latest American-designed processors have a strong incentive to optimize their AI models for less advanced hardware or to shift resources toward the “robot brain” software stack, where restrictions are harder to enforce and where the commercial payoff may be larger.
Washington Tightens the Semiconductor Chokepoint
The U.S. Department of Commerce, through its Bureau of Industry and Security, strengthened export controls to restrict China’s capability to produce advanced semiconductors for military applications. The package covers equipment, software, and high-bandwidth memory, adds firms to the Entity List, and includes compliance “red flag” guidance meant to help U.S. companies spot evasion attempts. The rules target the physical supply chain: the lithography tools, etching systems, and memory components that underpin chip fabrication.
Yet the controls leave a gap that matters for robotics. Humanoid robots need capable processors, but their competitive edge increasingly comes from the AI models running on those processors rather than from raw transistor counts. By restricting hardware while leaving software and model weights largely untouched, Washington may be channeling Chinese engineering talent toward exactly the layer where humanoid robots gain autonomy and commercial value. The BIS compliance framework was built for a semiconductor supply chain with identifiable choke points; AI model distribution does not have the same bottleneck structure, making enforcement far more difficult.
American Capital Finds Indirect Paths Into Chinese AI
Alongside hardware restrictions, the U.S. Treasury Department issued regulations implementing Executive Order 14105, establishing the Outbound Investment Security Program with an effective date of January 2, 2025. The program addresses U.S. investments in certain national security technologies and products in countries of concern, with sectors covered including AI. In theory, these rules should limit the flow of American dollars into Chinese AI ventures. In practice, the boundaries are porous.
U.S.-sourced limited-partner capital from pensions and endowments has funded China-facing AI venture activity through intermediary structures, according to reporting on Hong Kong financiers who specialize in routing institutional money into mainland startups. These intermediaries can sit outside the direct reach of some U.S. regulatory tools while still deploying American capital into sensitive sectors. The interaction between these capital flows and the new outbound-investment rules remains murky, especially because the Treasury regulations distinguish between prohibited transactions and those requiring only notification, leaving legal room for funds that frame their bets as commercial rather than strategic. For humanoid robot makers, that ambiguity can translate into continued access to deep pools of Western capital even as Washington tries to close the taps.
Robotics Strategy at the Edge of Policy
The tension between export controls, outbound-investment screening, and private capital is pushing Chinese humanoid robot companies to refine their strategic focus. With access to top-tier chips constrained, firms are emphasizing data-efficient training, model compression, and edge inference techniques that allow sophisticated control algorithms to run on mid-range processors. That shift aligns with the commercial realities of deploying thousands of robots into warehouses and factories, where cost, reliability, and power consumption matter as much as peak performance. In effect, the policy environment is nudging Chinese engineers toward practical, scalable AI for embodied systems rather than purely frontier research.
At the same time, the sheer scale of real-world deployment in China is giving local firms a structural advantage in the race to build general-purpose “robot brains.” Every humanoid robot shipped into a logistics center or industrial park can generate telemetry, video, and task data that feed back into centralized training pipelines. Over time, that data can support foundation models tuned specifically for manipulation, navigation, and human-robot interaction, differentiating Chinese platforms from rivals that rely more heavily on simulated environments. As long as capital, whether domestic or indirectly foreign, continues to underwrite these deployments, the feedback loop between hardware in the field and software in the lab will deepen, potentially outpacing what export controls alone can slow.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
