In brief
- ACE Robotics Chairman Wang Xiaogang predicts embodied AI could reach its “ChatGPT moment” by the end of 2027.
- A shortage of real-world training data remains a major obstacle, with ACE aiming to collect tens of millions of hours within two years.
- The Chinese startup plans to deploy its technology in 1,000 stores over the next year and eventually pursue an IPO.
Humanoid robots can walk, dance, and box, but getting them to perform useful work in the unpredictable physical world reliably remains one of artificial intelligence’s biggest challenges.
ACE Robotics Chairman Wang Xiaogang believes advances in AI models and real-world training data could soon give robots the intelligence needed to move beyond demonstrations and into commercial use, according to a report by Reuters.
“We expect to reach the ‘ChatGPT moment’ for embodied intelligence by the end of next year, driven by world models and environmental data capture,” Wang told Reuters.
Founded in July 2025, ACE Robotics is a Chinese startup developing AI models for humanoid robots. Backed by Ant Group and SenseTime, it raised more than $100 million in the first half of 2026 and, according to Reuters, plans to pursue an IPO “as early as permitted.”
While large language models such as ChatGPT and DeepSeek have spread rapidly, robots still struggle to perform a wide range of tasks in unfamiliar environments, and a lack of training data remains a major obstacle.
Embodied AI enables robots and other physical agents to perceive their environment, reason about it, and convert decisions into actions through sensors and actuators. At the same time, world models help AI understand how the physical world works by learning how objects and environments behave.
For robots, that means anticipating what might happen when they move, pick something up, or interact with their surroundings before taking an action.
“Over the past few years, the entire industry has accumulated data of roughly 100,000 hours, which is far from enough to train embodied foundation models,” Wang said.
Researchers elsewhere are experimenting with similar approaches. In October, researchers unveiled HumanoidExo, a wearable exoskeleton that captures human movements to train humanoid robots.
Other companies are developing their own AI models for robots.
In January, Boston Dynamics unveiled the production version of its Atlas humanoid, saying advances in AI helped bring the robot closer to commercial deployment. In June, Alibaba introduced its Qwen-Robot Suite, a set of AI models designed to help robots navigate, perform physical tasks, and simulate real-world environments.
Facts Only
ACE Robotics is a Chinese startup founded in July 2025.
Chairman Wang Xiaogang predicts a "ChatGPT moment" for embodied AI by the end of 2027.
ACE Robotics raised over $100 million in the first half of 2026.
Ant Group and SenseTime are backers of ACE Robotics.
The company intends to deploy technology in 1,000 stores within the next year.
ACE Robotics plans to pursue an IPO.
The industry has accumulated approximately 100,000 hours of training data.
ACE Robotics aims to collect tens of millions of hours of data within two years.
Boston Dynamics released a production version of the Atlas humanoid in January.
Alibaba introduced the Qwen-Robot Suite in June.
HumanoidExo is a wearable exoskeleton designed to capture human movements for robot training.
Executive Summary
The pursuit of embodied AI—the integration of AI models into physical agents capable of reasoning and action—is currently limited by a critical shortage of real-world training data. While large language models have scaled rapidly, humanoid robots struggle with reliability in unpredictable environments. To overcome this, developers are leveraging "world models" to help AI anticipate physical interactions and deploying hardware, such as wearable exoskeletons, to capture human movement data.
ACE Robotics, a Chinese startup backed by Ant Group and SenseTime, represents a significant commercial push in this sector. With over $100 million in funding, the firm aims to scale its presence to 1,000 stores and aggressively expand its dataset from the industry average of 100,000 hours to tens of millions. This effort mirrors broader industry movements, including Boston Dynamics' move toward commercial production of the Atlas robot and Alibaba's release of the Qwen-Robot Suite. While leadership at ACE predicts a breakthrough in intelligence by 2027, the timeline remains dependent on the successful acquisition and processing of massive amounts of environmental data.
Full Take
The strongest version of this narrative is that we are witnessing the transition of robotics from scripted demonstrations to generalized intelligence, mirroring the leap from basic chatbots to LLMs. The core argument is that data is the only remaining bottleneck; once the "data gap" is closed, commercial utility becomes inevitable.
However, this narrative relies heavily on the "scaling hypothesis"—the belief that more data automatically yields emergent intelligence. The projection of a "ChatGPT moment" by 2027 is a high-stakes claim that serves as a powerful signal to investors and IPO markets. Because the central premise—that intelligence will emerge if data reaches a specific threshold—rests entirely on the assertions of a company chairman seeking an IPO, the narrative functions as a projection of corporate ambition rather than a technical certainty.
Patterns detected: ARC-0043 Authority Game
The driving paradigm here is "Digital Determinism": the assumption that the physical world can be fully modeled and solved through data ingestion. This echoes the early 2020s AI hype cycles where "scale" was presented as a substitute for a fundamental understanding of cognition. The second-order consequence is a potential "gold rush" for physical data capture, where human movement and environmental interactions are commodified to feed corporate models.
If this were a coordinated influence campaign, the playbook would involve using "ChatGPT" as a linguistic anchor to trigger investor FOMO and using specific numerical targets (1,000 stores, millions of hours) to create an illusion of mathematical precision. The actual content aligns partially with this by framing a venture capital timeline as a technological prophecy, though it remains within the bounds of standard industry reporting.
Bridge Questions:
1. Is physical intelligence a matter of data volume, or does it require a fundamental architectural shift beyond current world models?
2. What are the privacy and ethical implications of collecting "tens of millions of hours" of real-world environmental data?
3. How does the "ChatGPT moment" framing bypass the unique safety risks inherent in physical, humanoid agents compared to text-based AI?
Sentinel — Human
The text appears to be a synthesis of real industry developments and expert predictions, exhibiting the characteristics of human-curated journalistic reporting rather than pure, uncontextualized machine generation.