ACE Robotics predicts a "ChatGPT moment" for robots: breakthrough slated for 2027
The embodied artificial intelligence industry stands on the brink of a tectonic shift. Based on my analysis of the latest trends and statements from key players, I estimate that by the end of 2027, we will witness a "ChatGPT moment" for robotics. This view is shared by Wang Xiaogang, chairman of the Chinese startup ACE Robotics, who links this breakthrough to the evolution of control algorithms and the exponential growth of data on physical interaction.
"We expect a 'ChatGPT moment' for embodied intelligence by the end of next year," Wang emphasized.
However, one should not be complacent: even after this technological leap, according to his forecasts, another four to five years will be needed for full commercialization. This is not about targeted demonstrations, but about integration into mass business processes.
The main bottleneck is not "hardware," but "brains"
Modern humanoids already impress with their agility—they run, dance, and perform highly specialized tasks. But their main Achilles' heel is the inability to adapt to unfamiliar environments and solve a wide range of tasks without prior programming. At ACE Robotics, they rightly believe that the problem lies not in the hardware, but in the immaturity of AI models.
This explains the development of their own model, Kairos, designed to combine perception, multimodal understanding, and real-time behavior planning. The results are already impressive: in July, Kairos topped four key tests for embodied AI, and its open version, Kairos-4B, despite a modest 4 billion parameters, outperformed giants like Nvidia Cosmos 3 and Ant Group's Lingbot in public benchmarks. This suggests that an efficient architecture can matter more than raw computing power.
Data scarcity—the industry's main challenge
The key constraint on the industry's development is the lack of high-quality data on real-world interaction. Over the past few years, the entire industry has accumulated only about 100,000 hours of such recordings—catastrophically insufficient for training foundation models.
Traditional data collection methods via teleoperation in exoskeletons are slow and expensive. ACE Robotics offers a more elegant solution: they equip ordinary workers on real production lines with lightweight sensors. This approach allows them to accumulate tens of millions of hours of data over two years, potentially giving the company a decisive competitive advantage.
Ambitious plans and geopolitical context
The startup is already testing its solutions in unmanned retail, hotel services, and warehouses. The plans are ambitious: to deploy the technology in 1,000 stores within a year, and in 10,000 by the end of the second. Notably, ACE Robotics is diversifying its chip supply, using both Nvidia accelerators and Chinese developments to reduce costs and dependence on a single supplier.
Since its founding in July 2025, the company has raised over $100 million from Ant Group and SenseTime, and plans to go public in the first half of 2026, as soon as local legislation permits. These plans align with the broader trend: Morgan Stanley analysts have already raised their forecast for Chinese humanoid shipments in 2026 from 28,000 to 50,000 units.
My analysis: ACE Robotics' success largely depends on whether they can scale data collection faster than competitors. If their strategy works, we could see not just evolution, but a true revolution in robotics that will reshape the labor market and logistics. However, betting on the Chinese market carries regulatory risks that could slow the pace of adoption.