From July 3rd to 4th, the 2026 Yabulu Forum Innovation Annual Conference was held in Shanghai. The theme of this year's conference was "Innovation-Driven, Intelligence Igniting the Future – The 15th Five-Year Plan Blueprint and Corporate Opportunities."

In his speech, Chen Li, co-founder of Unitree Robotics Co., Ltd., forecasted three core breakthroughs for the intelligent robotics industry within the next two to five years. First, a unified end-to-end robot large model, which he identified as the most crucial factor. Second, lower-cost, longer-lasting hardware coupled with ultra-large-scale mass production, which will significantly reduce marginal hardware costs. Drawing parallels with the automotive sector, he suggested that extreme cost-effectiveness after scaling could lead to rapid adoption. Third, low-cost, large-scale computing power. He proposed leveraging distributed computing to address challenges on the device side, such as battery life, heat dissipation, and cost, as well as latency issues in the cloud. For example, dedicated AI computing base stations in buildings could provide computational support for all on-site smart devices and robots.

He stated that the core indicator of the "GPT moment" for embodied intelligence is straightforward: achieving "two 80% thresholds." This means that in 80% of unfamiliar scenarios, a robot should be able to successfully complete approximately 80% of tasks based on voice or text instructions.

He further projected that by 2030, just four years from now, embodied intelligent robots will spearhead a new wave of consumer trends and significantly drive the advancement of social productivity and the labor force.