On 16 February 2026, the eve of the Year of the Horse, four Chinese humanoid robot manufacturers — Unitree Robotics, MagicLab, Galbot and Noetix Robotics — took to the biggest stage of the year, the CCTV Spring Festival Gala. The press focused on the choreography: Unitree’s G1s achieved several world firsts for a humanoid — the first continuous parkour table-vault, the first ejection backflip of more than three metres, the first continuous one-legged flips, and an Airflare of 7.5 rotations. Spectacular, and misleading. Because the news of the week is not the dance. It is what happens just after the curtain falls.
The stage was a test bench, not an end in itself
Wang Xingxing, founder of Unitree, did not sell a circus act. He sold an infrastructure. According to him, the cluster control performance at the Gala “lays the foundations for the future deployment of robots in cluster operations or single-unit deployment across other scenarios”. Translation: the stage was just a demonstration of simultaneous piloting. The proof followed swiftly. After the Chunwan, Unitree filmed 49 G1s performing a synchronised martial arts demonstration in front of the Temple of Heaven in Beijing — on the same cluster control platform as the one used at the gala. The TV studio and the public square share the same technical building block. The show was not the product; it was the commercial argument for the product.
The real figure is a purchase order
If one looks for the hard metric that the kung-fu folklore covers up, it is elsewhere. In the first two hours of the Gala broadcast, JD.com recorded a 300% rise in robot searches, a 460% rise in customer service requests and a 150% rise in order volume. And this demand was not confined to the metropolises: purchases came from more than a hundred Chinese cities, from first-tier agglomerations to small towns. The stage converted, in real time, viewers into buyers.
The industrial response is immediate and quantified. After the Chunwan, Unitree announced it was targeting the delivery of 20,000 humanoid robots in the year 2026, compared with approximately 5,500 units in 2025 — a fourfold increase in its manufacturing capacity, explicitly presented as a response to post-gala demand. The backflip is not the information. The x4 in capacity is.
On both sides of the Pacific, capital makes the same calculation
That same week, money confirms the diagnosis — and it no longer pays for the demo. On 11 February, Apptronik closed a Series A-X extension that brings its valuation to more than five billion dollars, approximately three times its initial 2025 Series A. The detail matters more than the amount: its robot Apollo is already deployed in real conditions, in factories and warehouses, at Mercedes-Benz and GXO Logistics. And the use of funds is clear — ramp up Apollo production, extend the global deployment network, and build facilities in Texas dedicated to robot training and data collection.
On the other side of the Pacific, the same calculation. Galaxea AI closed a Series B led by Jinding Capital, which lifts its post-money valuation to ten billion yuan, approximately 1.4 billion dollars. The composition of the round is the signal: three new industrial investors are entering — BAIC Group Industrial Investment, Hone Capital and Yifeng Capital. From Beijing to Austin, capital is no longer funding a choreography. It is funding the factory and the data.
The hidden motif: the human wearing the sensors
And the data, precisely, is still collected from a human. It is the most buried event of the week, and the most revealing. On the same day, or almost, that the G1s were dancing on television, AgiBot spun off Maniformer, a subsidiary in which it retains 75% of the capital — a controlled spin-off, not a divestment. Yet Maniformer’s product is not a robot. It is a data platform for embodied AI, positioning itself to fill a deficit estimated at more than five million hours of quality data in 2026.
Its method finally unmasks the sector: a “no physical robot” collection, where human operators wear sensor devices to perform tasks in real conditions and record trajectories, vision and tactile feedback. The human in the sensor harness is the robot, for the duration of the learning. And the operation is endorsed by capital: Sequoia China led its seed round, and the funds are earmarked for expanding data production capabilities, international deployment and the creation of a global data alliance for embodied AI.
Two sides of the same week
The “autonomous” robot of the Gala and the harnessed human of Maniformer are not opposed: they are the two sides of the same industry. One puts on the show and generates demand; the other quietly performs the learning that will make this demand sustainable. The question is therefore not whether the Gala robots were truly autonomous. It is who picks up the order placed on JD.com — and who wears the sensor harness meant to teach these machines to become so. The body dances, in the East as in the West. The data, however, is still collected from a human.