During the 2026 Yabuli Forum Summer Annual Meeting held in Chengdu from September 4th to 6th, themed "Corporate Innovation and Cycle Crossing," Yao Song, Founder and CEO of Zhengxing Innovation, shared his insights on the current state of embodied intelligence. He noted that this sector has drawn unprecedented attention, adding that the investment frenzy surrounding world models in the first half of this year could be the most extreme period many investors and institutions have witnessed in their decades-long careers.

Behind this surge, numerous companies have put forward a bold proposition: embodied intelligence, also known as physical intelligence, is highly likely to become the largest industry in human history. Yao Song referenced Elon Musk's prediction that by 2040, there will be 10 billion robots globally, with an average price of $20,000 to $25,000 per unit, ultimately surpassing the human population in numbers. The combination of a massive installed base and high unit prices is bound to create a wealth of new opportunities.

"I also believe that in the next decade, at least 40 to 50 new listed companies will emerge from this sector," he said. Despite the industry's current heat, Yao cautioned that as a founder, one must clearly recognize that embodied intelligence is a long-term endeavor. "The general-purpose embodied intelligence we all envision may truly take ten to twenty years to arrive. Deployment won't be as fast, and the race won't conclude as quickly as it has with large models," he explained.

He emphasized that in the field of embodied intelligence, true implementation requires the integration of three elements: models, hardware bodies, and scenario-based solutions. Merely having a model is not enough for direct application. The model might only account for about 20% of an entire project, even though it represents the most valuable portion. "Therefore, the complexity of embodied intelligence deployment, as well as the difficulty of building industry barriers, will certainly be higher than that of large models, and the competitive landscape won't solidify rapidly," he concluded.