The year 2025 has become a breakout year for AI drug discovery in China. Companies such as YuanSi Biopeptides, HuaShen AI Pharma, and XtalPi have completed business development (BD) deals worth several billion USD in total. While startups are achieving dazzling results, major players are also heavily investing in the AI drug discovery space: in September, China Telecom launched a “Public Service Platform for AI Drug Discovery,” and ByteDance announced the successful experimental validation of its PXDesign system.
1. 2025: The Year AI Drug Discovery Explodes
2025 has seen multiple landmark events in AI drug discovery, reflecting the rapid development of the field.
Major Collaboration Projects:
- XtalPi announced a strategic partnership with DoveTree with a potential total value of USD 5.99 billion. The upfront payment alone reached USD 51 million, proving that AI drug discovery platforms are already capable of generating real cash flow.
- In early October, Algen Biotechnologies announced a multi-target collaboration with AstraZeneca to jointly advance AI-powered drug discovery in immunology, with a potential deal value of USD 555 million.
- China Telecom, in collaboration with Bayer, Hengrui, IQVIA, and other global pharmaceutical giants, officially launched the “Public AI Drug Discovery Service Platform,” signaling its formal entry into the field and the formation of a national-level AI drug discovery team.
Technological Breakthroughs:
- By mid-2025, multiple institutions released large biological models with de novo (from-scratch) design capabilities. Chai Discovery, backed by OpenAI, launched the Chai-2 model, boosting antibody design success rates from the traditional 0.1% to 15.5%. The development cycle was shortened from several months or even years to just two weeks, and development costs were also drastically reduced.
- ByteDance announced a breakthrough in AI drug discovery: the PXDesign system achieved nanomolar-level binding hit rates between 20% and 73% on 5 out of 6 protein targets. In comparison, DeepMind’s AlphaProteo, based on its AlphaFold models, achieved only 9%–33% hit rates on the same targets.
Regulatory and Clinical Progress:
- An increasing number of AI-driven drug pipelines are entering Phase II and Phase III clinical trials. The resulting trial data will become a crucial benchmark for validating the industry’s real-world value. For example, Chinese company Insilico Medicine’s leading pipeline product NouvNeu001 has already entered a multi-center Phase II clinical trial.
2. Industry Progress: A Fruit of Long-Term Accumulation
AI drug discovery didn’t rise overnight; it went through several stages of accumulation and breakthrough.
- Exploratory Phase (before 2010):
AI’s application in drug discovery remained within the realm of traditional computer-aided drug design. The pharmaceutical industry still primarily relied on trial-and-error R&D models. - Technology Accumulation and Verification Phase (2010–2016):
With breakthroughs in machine learning and deep learning algorithms, AI began to be used in target discovery and molecular screening. Founded in 2012, Atomwise became a representative company during this period. During the 2015 Ebola outbreak, it predicted two potentially effective drugs within one week using its AI platform, demonstrating the potential of AI in drug discovery. - Rapid Development Phase (2016–2020):
Marked by DeepMind’s AlphaGo and AlphaFold, AI drug discovery began receiving wide attention. In 2020, DeepMind released AlphaFold2, which successfully predicted protein structures, greatly advancing applications of AI in structural biology. - Commercialization Phase (2020–present):
AI drug discovery has shifted from technological validation to industrialization and commercialization. More AI pharma firms are forming partnerships with large pharmaceutical companies, and some AI-designed candidate drugs have already entered clinical trials.
AI is now deeply applied across multiple stages of drug development:
- Target Discovery & Validation:
AI analyzes genomic and other biomedical data to identify new disease targets and accelerates protein structure modeling and validation. Representative platforms include Insilico Medicine’s PHARMA.AI and XtalPi. - Molecule Design & Optimization:
AI generates and screens candidate molecules with desired bioactivity and predicts ADMET properties (Absorption, Distribution, Metabolism, Excretion, and Toxicity). Examples include ChengYu Biotech’s “Mol Prospector” and Tencent AI Lab’s IgGM model. - Complex Drug Design:
For designing antibody-drug conjugates (ADC) and other complex therapies, AI helps optimize antibody, linker, and payload combinations. Representative product: TherapiAI’s ADC AI Agent. - Preclinical Testing:
AI assists in designing automated synthesis routes and high-throughput experimental validation. Examples include ZhiHua Tech’s AI Chemist platform and XtalPi’s automated lab.
Why AI Drug Discovery Exploded in 2025:
- Technological Maturity:
The latest AI boom stems from major breakthroughs in general-purpose AI, whose capabilities are now spilling over into drug R&D and enzyme design. Explosive growth in biomedical data has also provided ample training material. - Protein Prediction & Design Breakthroughs:
AlphaFold2 confirmed the effectiveness of Transformer architectures in decoding the “language of life.” David Baker’s team applied Diffusion models from image generation to biology, dramatically improving success rates in novel protein design. - Efficiency Revolution:
Traditional antibody drug R&D could take 3 years and cost USD 5 million from target validation to screening effective antibodies. Now, AI models like Chai-2 can complete the process in hours and validate via wet-lab experiments within two weeks. - Capital Push:
Since 2024, financing activity in global AI drug discovery has surged. Funding in H1 2024 already surpassed the total for 2023. Tech giants like NVIDIA and OpenAI have invested deeply via equity or technical platforms. - Policy Support:
China’s “14th Five-Year Plan for the Pharmaceutical Industry” explicitly encourages the application of AI, cloud computing, and big data in R&D. Local governments in Beijing, Shanghai, Jiangsu, and Zhejiang have also rolled out supportive policies.
3. The Irreversible Trend: AI Will Become the Default Tool for Drug R&D
The future of AI-powered drug development is unfolding in several clear directions:
- Full-Process Integration:
AI will expand from early discovery into clinical trial design and manufacturing. DeepWisdom’s generative AI platform can already auto-generate clinical trial protocols and has passed Japan PMDA’s review on the first try—showcasing deep optimization potential. - Deep Tech Integration:
Integration of multimodal data and breakthroughs in generative AI (e.g., GANs, Transformers) will push AI drug discovery beyond template-based design into de novo molecule design. - Industry Restructuring:
In the future, AI-powered biotech companies may become “molecular design centers” and “computational hubs” for global pharma companies, focusing on high-frequency, high-tech discovery. Large pharma will focus more on clinical trials, regulatory approvals, and commercialization. - Breakthroughs in Chronic Diseases:
AI will dramatically shorten preclinical development cycles for oncology, autoimmune, and metabolic diseases. “Chronic conditions” are likely to benefit first, increasing the chance of discovering blockbuster drugs like Semaglutide. - Personalized Medicine:
Integration of AI and multidisciplinary technologies will drive precision medicine. Real-time AI platforms deployed in trials will dynamically adjust treatments—becoming a future development focus. - Evolving Regulatory Frameworks:
Global regulators are actively exploring AI compliance in drug development. The U.S. FDA published draft guidelines in 2023 on “AI/ML in Drug Development,” and the European EMA launched an “AI in Medicine” initiative supporting AI-driven pharmacovigilance systems.
Conclusion
AI drug discovery is no longer just a buzzword—it’s fundamentally reshaping the logic of drug development: from the traditional “screen what you have” to the AI-driven “build what you want.” This shift not only increases R&D efficiency but may also overcome previously untreatable diseases, bringing new hope to patients.
As AI drug discovery technologies mature, industry standards will gradually be established, and technical barriers will rise. At the current pace, every new drug R&D company will eventually use AI.
[Disclaimer]: The above content reflects analysis of publicly available information, expert insights, and BCC research. It does not constitute investment advice. BCC is not responsible for any losses resulting from reliance on the views expressed herein. Investors should exercise caution.