Chinese companies building artificial intelligence for robots raised 93.5 billion yuan ($13 billion) across 322 funding deals in the first half of 2026, five times the amount raised in the same period a year earlier, according to Reuters. The surge marks a shift in China's robotics industry away from building increasingly agile robot bodies and toward the software that decides what those bodies actually do.
Startups including X Square, Galaxea and Spirit AI have each reached roughly $3 billion valuations on the strength of so-called embodied AI: systems that let a robot interpret its surroundings, pick up new tasks and adapt without being reprogrammed for every job. Spirit AI co-founder Gao Yang told Reuters the brain remains "the weakest link in the complete robotics stack," while X Square co-founder Yang Qian said the goal is for robots to be "driven by common sense" rather than hand-written code.
Why the money is moving from bodies to brains
Years of fierce competition among Chinese humanoid and quadruped makers have pushed hardware prices down, leaving less room for profit on the robot body itself. Dermot McGrath of ZenGen Labs told Reuters that manufacturers have concluded "the margin won't come from the hardware but from the brain," a calculation redirecting investor cash toward perception and decision-making software rather than actuators and frames.
That shift shows up in how companies gather training data. Spirit AI has hired roughly 1,000 people to record everyday and industrial activities for its models, while Ant Group's robotics unit, Robbyant, said it trained a system on 160,000 hours of material, including 50,000 hours of human-guided robot movements collected across 20 robot configurations from 17 hardware brands. DexForce, which builds simulated environments for training, said some of its industrial tasks reach deployment within six to eight hours and achieve success rates above 99.9% in precision operations. Battery maker CATL reported that Spirit AI's robots exceeded 99% success inserting flexible wires during testing at a plant in Zhongzhou.
The data-hungry nature of the work is also drawing in companies outside traditional robotics. ACE Robotics is using wearable sensors to capture human movement for training sets, while spatial-intelligence firm Manycore Tech and JoyIn Technology, which is developing an AI system called Aether, are building complementary layers of the stack. Progress on general tasks still remains modest: Generalist's GEN-1.5 model completed simple tasks at only a 60% success rate after three to twelve seconds of human demonstration, a reminder of how far embodied AI has to go before it matches narrow, single-task automation.
A harder constraint than money
Even with capital flowing freely, Chinese embodied-AI developers face a bottleneck that funding cannot fix on its own: US export controls have limited access to Nvidia's most advanced chips, slowing the computing-heavy work of training large models. Companies are turning to domestic alternatives from Huawei and other chipmakers, though it is unclear whether those substitutes can match Nvidia's performance for the largest training runs. UBTech, a Hong Kong-listed humanoid maker, has advertised a chief AI scientist role paying up to 124 million yuan ($18 million), a sign of how aggressively firms are bidding for scarce technical talent even as hardware budgets stay tight.
The pivot also reflects a harder lesson from factory and logistics deployments: industrial customers have little tolerance for improvisation. Suppliers face long validation cycles and must integrate their systems site by site before robots can be trusted to run unsupervised on a production line, pushing developers to prioritize reliability over flashy demonstrations.
The funding total puts China's embodied-AI sector on a pace that dwarfs most Western robotics-software rounds disclosed so far in 2026, though valuations above $3 billion remain concentrated in a handful of firms. Whether that capital translates into robots that can reliably handle unscripted tasks outside controlled factory settings is the question investors are now betting billions of yuan will be answered within the next year.