The exhaustive trial-and-error approach struggles to perform a global optimal solution in increasingly complex and high-dimensional chemical research. Artificial intelligence (AI), with its ability to decipher the high-dimensional correlations in complex data, offers a transformative opportunity to address this challenge. The automated experimental platform can produce sufficiently standardized data for intelligent models by executing high-throughput experimentation, and equip these models with practical research capabilities. By combining AI with automated experiments, the AI-Chem team at University of Science and Technology of China has developed the first all-around and data-intelligence-driven robotic AI chemist in the world. Using predictive models based on theoretical computations and experimental feedback, the robotic AI chemist has achieved the global optimal synthetic formula, along with a series of benchmark examples for the new data-driven research paradigm. To further promote this paradigm, a cloud infrastructure for AI chemist is proposed, which includes multi-domain, multi-modal, and standardized intelligent databases; large models based on mathematical logic; distributed and fully automated intelligent experimental facilities; and an intelligent chemistry cloud platform for resource allocation and sharing. This cloud infrastructure can guide a new research paradigm that combines intelligence and materials science, thereby propelling China’s chemistry discipline to the forefront of the international stage.