AI for Science has become more arresting since the 2024 Nobel Prizes for Physics and Chemistry were awarded for AI-related achievements. With Zhongguancun, "China's Silicon Valley," Beijing aims to lead the integrated development of AI and science, especially in the new materials sector.

In January, an action plan to accelerate innovative development of AI + new materials was jointly issued by the Beijing Municipal Science and Technology Commission,  Administrative Commission of Zhongguancun Science Park, the Beijing Municipal Bureau of Economy and Information Technology, and the Beijing Municipal Commission of Development and Reform.

As per the plan, by 2027, Beijing's innovation capability in AI + new materials will be significantly strengthened, with progress in the cultivation of the new materials development service industry. Beijing will become a world-leading source of new materials innovation and a highland for AI application.

The current number of AI + new materials innovative enterprises in Beijing makes up one-third of the national total, with outstanding achievements in AI empowering new materials development, according to Zhou Hang, deputy director of the Beijing Advanced Materials and Energy Technology Development Center.

The China Iron and Steel Research Institute Group Co. developed a high-performance hydrogen-resistant thick plate material with the world's highest strength (800MPa level) in six months. The AI for Science Institute, Beijing, developed a new-generation OLED (organic light emitting diode) luminescent material without any precious metals with the help of a molecular representation learning model.

The main advantages of using AI to develop new materials are speedy development of new molecules, greatly advancing the existing technology, and enhancing the efficiency of experiments.

Under the Large Atom Model Initiative, the next-generation Uni-Mol molecular model by DP Technology, is second only to Google's AlphaFold3 in its ability to protein structure prediction accuracy, but its development cost is a fraction of a percent of the latter's, according to Wu Xu, senior manager of solutions at DP Technology.

Zhu Shucun, manager at the engineering department of DeepChem, said their high-throughput experiment platform, part of a smart synthesizing system, can conduct experiments 24/7 far more efficiently than artificial experiments. The platform can also do whole-process automation and monitoring, and the data generated is transmitted to the database directly.

The system integrates with external LLMs to meet the needs of different sectors and enterprises, according to Liu Yuzhou, founder and CEO of DeepChem and a professor at Beihang University.

Besides DP Technology and DeepChem, many other enterprises in Beijing are accelerating the construction of intelligent laboratories for new material development.

Zhou said Beijing has been assigned several tasks for building the national big data center for new materials. It is leading the construction of the service portal of the center's main platform. The portal's official version is expected to be launched in December.  

Three of the four tasks in the data resource nodes initiated in 2024 are being carried out in Beijing, and more industrial data resource nodes are expected to be built by the capital, laying a solid foundation for the data-driven, intelligent and innovative development of new materials, Zhou said.