Wang Xingxing: Embodied AI Set for ChatGPT-Like Breakthrough in 2-3 Years

Wang Xingxing, founder and CEO of BotSchool, highlighted during his address at the 2026 World Robot Conference (WRC2026) main forum that embodied intelligence is nearing an industrial inflection point akin to the "ChatGPT moment." He projected that breakthroughs could occur within as little as two to three years, or potentially take five to ten years under a slower adoption scenario.
BotSchool’s listing on the Shanghai Stock Exchange’s Science and Technology Innovation Board occurred just one day prior to Wang’s speech, with the stock closing 460% above its issue price on its debut trading day. Wang emphasized that the true indicator of embodied intelligence entering an industrial boom is not isolated demonstrations, but the ability of robots to execute approximately 80% of tasks via voice or text commands in 80% of unfamiliar environments.
According to Wang, the primary technical hurdle for humanoid robots remains aligning AI models with physical reality. While robots perform well in fixed, fully trained scenarios, their success rates plummet when objects or environments shift slightly. Challenges in tactile feedback and precision errors during the final centimeters or millimeters of operation continue to restrict generalization capabilities.
To tackle these challenges, BotSchool unveiled its pre-research roadmap for "Physical AI Robot Self-Evolution V1.0." This system leverages large AI models to automatically retrieve research, generate control code, and establish a continuous iteration loop through simulation, real-world testing, and combined AI-human evaluation. BotSchool identifies foundational model capabilities, multi-source data accumulation, real robot deployment scale, and skill accumulation as key drivers for this self-evolution.
Wang also traced BotSchool’s evolution from quadruped to humanoid robots, noting that their G1 humanoid has become an industry benchmark, featuring in projects like the Spring Festival Gala’s "WuBOT." This year, BotSchool introduced additional products, including a manned exoskeleton and the wheeled-legged robot As2-W, while continuing to explore applications across factory, home, and outdoor settings.
Wang noted that achieving large-scale deployment requires resolving efficiency and generalization issues. Although robots can handle simple tasks, new tasks often demand retraining. However, as AI models improve and real-world data accumulates, embodied intelligence is poised to accelerate its transition into industrialization.
As humanoid robots transition from experimental demos to commercial use, data, models, hardware, and self-evolution capabilities are emerging as core competitive factors. BotSchool’s roadmap reflects a broader industry shift from merely "manufacturing robots" to "enabling robots to learn and evolve continuously."
Related article
Preply’s Unicorn Status Reflects Ukrainian Resilience
Language learning platform Preply has reached a $1.2 billion valuation following a $150 million Series D funding round, marking a significant milestone for the 14-year-old company. Previous investors include Horizon Capital, Hoxton Ventures, Owl Vent
Microsoft revives in-house AI strategy with new coding model as Claude costs soar
AI-assisted coding has become a staple in modern software development, yet even tech giants like Microsoft are feeling the pinch of expensive third-party large language model subscriptions. To cut external dependencies and operational overhead, Micro
ByteDance and HKUST Unveil MMProLong to Boost Long Document LMM Training
On May 24, the ByteDance Seed team partnered with the Hong Kong University of Science and Technology to unveil a breakthrough in long-document training for multimodal large language models (LMMs). Leveraging Alibaba’s open-source Qwen2.5-VL, the team
Related Special Topic Recommendations
Comments (0)
0/500

Wang Xingxing, founder and CEO of BotSchool, highlighted during his address at the 2026 World Robot Conference (WRC2026) main forum that embodied intelligence is nearing an industrial inflection point akin to the "ChatGPT moment." He projected that breakthroughs could occur within as little as two to three years, or potentially take five to ten years under a slower adoption scenario.
BotSchool’s listing on the Shanghai Stock Exchange’s Science and Technology Innovation Board occurred just one day prior to Wang’s speech, with the stock closing 460% above its issue price on its debut trading day. Wang emphasized that the true indicator of embodied intelligence entering an industrial boom is not isolated demonstrations, but the ability of robots to execute approximately 80% of tasks via voice or text commands in 80% of unfamiliar environments.
According to Wang, the primary technical hurdle for humanoid robots remains aligning AI models with physical reality. While robots perform well in fixed, fully trained scenarios, their success rates plummet when objects or environments shift slightly. Challenges in tactile feedback and precision errors during the final centimeters or millimeters of operation continue to restrict generalization capabilities.
To tackle these challenges, BotSchool unveiled its pre-research roadmap for "Physical AI Robot Self-Evolution V1.0." This system leverages large AI models to automatically retrieve research, generate control code, and establish a continuous iteration loop through simulation, real-world testing, and combined AI-human evaluation. BotSchool identifies foundational model capabilities, multi-source data accumulation, real robot deployment scale, and skill accumulation as key drivers for this self-evolution.
Wang also traced BotSchool’s evolution from quadruped to humanoid robots, noting that their G1 humanoid has become an industry benchmark, featuring in projects like the Spring Festival Gala’s "WuBOT." This year, BotSchool introduced additional products, including a manned exoskeleton and the wheeled-legged robot As2-W, while continuing to explore applications across factory, home, and outdoor settings.
Wang noted that achieving large-scale deployment requires resolving efficiency and generalization issues. Although robots can handle simple tasks, new tasks often demand retraining. However, as AI models improve and real-world data accumulates, embodied intelligence is poised to accelerate its transition into industrialization.
As humanoid robots transition from experimental demos to commercial use, data, models, hardware, and self-evolution capabilities are emerging as core competitive factors. BotSchool’s roadmap reflects a broader industry shift from merely "manufacturing robots" to "enabling robots to learn and evolve continuously."
Preply’s Unicorn Status Reflects Ukrainian Resilience
Language learning platform Preply has reached a $1.2 billion valuation following a $150 million Series D funding round, marking a significant milestone for the 14-year-old company. Previous investors include Horizon Capital, Hoxton Ventures, Owl Vent
Microsoft revives in-house AI strategy with new coding model as Claude costs soar
AI-assisted coding has become a staple in modern software development, yet even tech giants like Microsoft are feeling the pinch of expensive third-party large language model subscriptions. To cut external dependencies and operational overhead, Micro
ByteDance and HKUST Unveil MMProLong to Boost Long Document LMM Training
On May 24, the ByteDance Seed team partnered with the Hong Kong University of Science and Technology to unveil a breakthrough in long-document training for multimodal large language models (LMMs). Leveraging Alibaba’s open-source Qwen2.5-VL, the team





Home






