Key Takeaways
- • On October 8, 2026, the "X-Day" Xilihu Roadshow Society and the "Venture Capital China" Sci-Tech Roadshow Hall hosted an AI for Science (AI4S) special session at Tsinghua University that drew more than 200 representatives from government, university research teams, venture capital, industrial capital and industrial service providers.
- • Investors on the panel described AI4S as still early-stage, with data scarcity, the dry-wet lab closed loop and model prediction accuracy in vertical domains cited as the main hurdles.
- • Pharmaceuticals were identified as the track with the clearest near-term commercial value, monetized through phased transactions tied to clinical stages, while materials commercialization was viewed as harder given a single paying subject and many industrial constraints.
- • Panelists expect only a limited number of AI4S companies to grow into large listed firms, with the remainder likely to be acquired or to exit through partnerships with large manufacturers.
- • Six roadshow projects spanning scientific computing power, AI chips, material agents, autonomous laboratories, virtual cells and virtual organoids were presented at the session.
Summary
On the afternoon of October 8, 2026, the "X-Day" Xilihu Roadshow Society joined with the "Venture Capital China" Sci-Tech Roadshow Hall to bring an AI for Science special session to Tsinghua University, according to the publication's report. The event gathered more than 200 representatives from government authorities, university research teams, leading venture capital institutions, industrial capital and industrial service providers. Attendees included Liu Minqiang of the Investment Promotion Agency of the Ministry of Commerce, Li Zhina, Vice District Mayor of Nanshan District, Shenzhen, Peng Gang, Vice President of Tsinghua University, and venture investors including Ni Zhendong of Zero2IPO Group, Li Zhu of Inno Fund, Hua Hua of Huakong Fund, Ji Wei of China Renaissance Capital, Wang Xuehui of Shuimu Alumni Fund and Liu Bo of TusStar Venture Capital.
The report describes AI for Science as the fifth major scientific research paradigm after experiment, theory, computation and data. Supported by relevant policies in Beijing and the build-out of the Zhongguancun (Haidian) Scientific Intelligence Innovation Cluster, AI4S has moved from academic research to a national strategic direction, with applications in new materials, biomedicine and quantum information. Hua Hua of Huakong Fund cited three challenges: improving prediction accuracy of vertical domain models built on large models, forming a complete dry and wet laboratory closed loop, and insufficient data reliability and "AI readiness." He said the field remains early-stage and emphasized founding team capability as a core investment criterion.
Ji Wei of China Renaissance Capital contrasted materials with pharmaceuticals. Materials demand is clear, but the value of AI R&D is harder to demonstrate in end products, with a single paying customer base and many industrial constraints. Pharmaceuticals benefit from a mature division of labor and asset transaction system, allowing monetization through phased payments at clinical stages. She also noted a significant "island effect" in industry data, with high-quality trainable data scattered across laboratories and industrial operators.
Li Zhu of Inno Fund said AI4S is among the few high-quality venture tracks still growing amid a broader contraction in venture capital. He cited the compression of material R&D cycles from roughly ten years to one year. Wang Xuehui of Shuimu Alumni Fund said mathematics and physics mainly demonstrate model capability with limited near-term profitability, while pharmaceuticals and materials offer more implementation paths. He estimated that few companies in the track will become large listed firms, with others likely to be merged or to exit through cooperation with large manufacturers.
In a closing remark, Liu Bo of TusStar Venture Capital said AI4S is being applied across pharmaceuticals, materials, industrial manufacturing, quantum and nuclear energy, and highlighted interdisciplinary breakthroughs such as the interpretability of traditional Chinese medicine. Six roadshow projects were presented, including Quantum Kun Technology (quantum-AI-HPC scientific computing), Xinyi Technology (RRAM-based AI inference chips) and He Guang Zhi Cheng (a material design agent and autonomous laboratory model).
The report describes AI for Science as the fifth major scientific research paradigm after experiment, theory, computation and data. Supported by relevant policies in Beijing and the build-out of the Zhongguancun (Haidian) Scientific Intelligence Innovation Cluster, AI4S has moved from academic research to a national strategic direction, with applications in new materials, biomedicine and quantum information. Hua Hua of Huakong Fund cited three challenges: improving prediction accuracy of vertical domain models built on large models, forming a complete dry and wet laboratory closed loop, and insufficient data reliability and "AI readiness." He said the field remains early-stage and emphasized founding team capability as a core investment criterion.
Ji Wei of China Renaissance Capital contrasted materials with pharmaceuticals. Materials demand is clear, but the value of AI R&D is harder to demonstrate in end products, with a single paying customer base and many industrial constraints. Pharmaceuticals benefit from a mature division of labor and asset transaction system, allowing monetization through phased payments at clinical stages. She also noted a significant "island effect" in industry data, with high-quality trainable data scattered across laboratories and industrial operators.
Li Zhu of Inno Fund said AI4S is among the few high-quality venture tracks still growing amid a broader contraction in venture capital. He cited the compression of material R&D cycles from roughly ten years to one year. Wang Xuehui of Shuimu Alumni Fund said mathematics and physics mainly demonstrate model capability with limited near-term profitability, while pharmaceuticals and materials offer more implementation paths. He estimated that few companies in the track will become large listed firms, with others likely to be merged or to exit through cooperation with large manufacturers.
In a closing remark, Liu Bo of TusStar Venture Capital said AI4S is being applied across pharmaceuticals, materials, industrial manufacturing, quantum and nuclear energy, and highlighted interdisciplinary breakthroughs such as the interpretability of traditional Chinese medicine. Six roadshow projects were presented, including Quantum Kun Technology (quantum-AI-HPC scientific computing), Xinyi Technology (RRAM-based AI inference chips) and He Guang Zhi Cheng (a material design agent and autonomous laboratory model).
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.