Listed on the stock exchange at $66 billion, the Chinese robotics star has seen its share price fall by almost half. Analysts point to the same flaw: robots move well, but do not yet produce value.
Unitree's entry into the Chinese market had valued the manufacturer at $66 billion, a record for a robot manufacturer. The following week, the title had lost almost half. The correction is not an isolated market accident: it reflects a fundamental doubt about the ability of current machines to perform a truly profitable job.
The dominant diagnosis in the sector is that of a delay in software maturity. Harry Mellsop of Antioch places physical robotics in its "GPT-2 era": architectures work, but there is still a lack of data and computing power to cross the next level. Autonomous vehicles are an exception, precisely because they have long had volumes of driving data that the handling of objects has never had.
The precise tasks advance, the generalist humanoids stagnate
There is a clear contrast between companies targeting a defined profession and those promising a universal robot. Gritt mounts solar farms, Agility deploys its machines in the industry, Bedrock makes excavators dig without operator: all come out of the laboratory. Multipurpose humanoids rarely emerge. However, capital continues to flow — Genesis AI raised $105 million this year — and the current 2026 conference brought together 1,500 participants, three times more than in 2023.
Read the full article on the original source
Go to the source →