Shares of Unitree Robotics have fallen 45% from their IPO peak, erasing $30 billion in market value within days of their debut. This sharp correction highlights a growing reality check in the physical AI sector, where investors are now shifting focus from speculative market valuations to sustainable profits and commercial delivery.
The initial excitement surrounding the physical AI and humanoid robotics sector has met a harsh reality check. Unitree Robotics, which debuted on the Shanghai Stock Exchange’s STAR Market on August 19, 2026, saw its market value surge to a peak of approximately $66 billion on its first day of trading. However, investor enthusiasm quickly faded, and by August 25, 2026, the company’s share price had dropped by 45% from its IPO peak, wiping out roughly $30 billion in market capitalisation.
This stock movement signals a broader change in sentiment. Investors are increasingly questioning whether the extremely high valuations for robotics firms match their current financial performance and commercial capability. For instance, Unitree reported a 53% year-over-year decline in its adjusted net profit for the first quarter of 2026, a financial reality that contrasts sharply with the initial hype surrounding its public offering.
The broader physical AI sector, which saw $47.4 billion in global venture funding during the first half of 2026, is currently facing a 'valuation bubble' risk. Analysts note that while the technology is exciting, developers are struggling to scale from laboratory prototypes to mass-market commercial products. Unlike software, which can be deployed instantly, hardware requires expensive manufacturing, physical integration, and complex testing, creating a 'physical gap' that keeps costs high and production slow.
A significant hurdle identified by industry experts is the scarcity of high-quality data. While AI models like ChatGPT were trained on vast amounts of internet text, robots require complex sensory data—including visual, kinetic, and lidar information—that is both difficult and costly to collect. This 'data bottleneck' remains a primary obstacle for firms attempting to move beyond experimental research.
Moving forward, the debate among industry players is whether to focus on narrow, task-specific robotics—such as for industrial cleaning or solar farm maintenance—to generate immediate revenue, or to keep chasing the 'universal' humanoid model. Investors are likely to track whether companies can bridge the gap between technological potential and real-world utility. The key monitorable for the sector will be tangible signs of commercial delivery, such as stable production timelines and consistent profit growth, rather than just market hype.