China Securities Co.,Ltd.: AI reshapes the underlying logic of pharmaceuticals, accelerating breakthroughs in the commercialization validation phase

China Securities Co.,Ltd. released a research report stating that AI is reshaping the underlying logic of pharmaceuticals and accelerating breakthroughs in commercialization validation. Traditional pharmaceutical development faces the "double ten" dilemma of high investment, long cycles, and low success rates. AI technology has now fully penetrated core pharmaceutical processes such as target identification, virtual screening, de novo design, ADMET prediction, and automated synthesis, effectively overcoming the throughput limits of traditional high-throughput physical blind screening and greatly improving pharmaceutical efficiency and success rates. Currently, the AI pharmaceutical industry chain has evolved into a network-based enabling system centered on platforms and technology middle offices, with deep collaboration across upstream, midstream, and downstream segments, and is showing a development trend in which multiple business models advance in parallel. The global AI pharmaceutical market was approximately US$2.49 billion in 2025 and is expected to exceed US$46 billion by 2035, with a CAGR likely exceeding 33%. China's market growth rate is in sync with the global market and is expected to become one of the most certain incremental markets. China Securities Co.,Ltd.'s main views are as follows: AI is accelerating the reshaping of traditional pharmaceutical R&D paradigms. Drug action is essentially the interaction between molecules and targets in the body, and traditional drug R&D faces the underlying pain point of Eroom's Law. AI technology has now fully penetrated core processes such as target identification, virtual screening, de novo design, ADMET prediction, and automated synthesis, effectively overcoming the throughput limits of traditional high-throughput physical blind screening in pharmaceutical R&D and greatly improving pharmaceutical efficiency and success rates. AI pharmaceutical business models are accelerating their evolution, with full-stack enabling maps advancing in coordination. Upstream computing infrastructure is upgrading toward technology solutions, while midstream full-stack service providers are pursuing multiple lines simultaneously. Four major tierstech giants, BioTechs, traditional pharmaceutical companies, and CXOsare actively positioning themselves, and the integration of software services and self-developed pipelines has become the mainstream business model. As technology dividends are realized, the global and Chinese AI pharmaceutical markets are entering a period of rapid growth, and China, with its strong industrial resilience, is expected to become the most certain global incremental engine. A number of benchmark companies with platform capabilities and commercialization validation have emerged in global AI pharmaceuticals. Google laid the foundation for structure prediction with the AlphaFold series, the IsoDDE engine opened the prelude to commercialization, and Isomorphic Labs completed US$2.1 billion in financing; Anthropic acquired Coefficient Bio for US$400 million to complete full-stack drug R&D capabilities; Schrdinger built an industrial-grade computing platform driven by the dual wheels of "physics + AI"; Recursion connected the dry-wet experiment loop and integrated more than 50 PB of proprietary data; INSILICO advanced pipelines across multiple fields relying on the Pharma.AI platform, and with recent multi-point cooperation with giants such as Eli Lilly, cumulative BD contract value has reached nearly US$7.5 billion; XTALPI achieved a business explosion through its R&D flywheel of "AI + Siasun Robot&Automation" and is accelerating expansion into materials, consumer, and other fields; JITAI Technology built NanoForge, the world's first AI nanodelivery platform, with its core pipeline already advanced to the pre-NDA stage. AI pharmaceutical companies are generally in a stage of high R&D investment and not yet profitable, making traditional PE/PEG valuation methods difficult to apply. A sum-of-the-parts valuation framework is recommended: 1) For self-developed pipelines, use the DCF/FCFF method, deriving fair enterprise value through the three steps of "peak sales estimationsingle-pipeline cash flow modelingprobability-weighted discounting"; 2) For AI+CRO business, model upfront payments, milestone rewards at each stage, and long-term sales royalties separately, then sum and discount them; 3) For software licensing/solutions, refer to value data-driven platform companies and, comprehensively considering the scarcity, high growth, and horizontal expansion potential of AI pharmaceutical platforms, apply the PS method for valuation. Risk warnings: 1. Risk of technology translation falling short of expectations. AI pharmaceuticals are still in a critical window period moving from algorithm validation to clinical realization, and the generalization ability of models in complex biological systems still needs to be verified. If AI-driven candidate molecules fail to show significant advantages over traditional paths in the clinical stage, the pace of overall industry value re-rating may slow somewhat. 2. Risk that the commercial closed loop remains to be verified. Although the integrated model of "software + services + pipelines" is the industry's consensus direction, a complete profit closed loop is still being explored. Some companies are highly dependent on collaboration milestone income and licensing payments, and if the pace of external cooperation fluctuates periodically, short-term cash flow may come under certain pressure. 3. Risk of constraints from computing power and data elements. AI pharmaceuticals rely heavily on high-quality biomedical data and large-scale computing resources. Against the backdrop of adjustments in the global supply chain landscape, the availability of upstream computing resources and the compliant acquisition cost of high-quality data may both change marginally, and their potential impact on corporate R&D efficiency needs attention. 4. Risk of evolving competitive landscape in the sector. As tech giants, traditional pharmaceutical companies, CXOs, and emerging TechBio enterprises all increase their positioning efforts, industry participants are becoming increasingly diverse. Against the backdrop of intensifying competition, the difficulty of building differentiated positioning and core barriers has risen, and pricing space in some sub-segments may face a certain degree of narrowing pressure. 5. Risk of dynamic adjustments in regulatory policies. Globally, the regulatory framework for AI-assisted drug R&D is still in a stage of gradual improvement, and policy directions regarding algorithm interpretability, cross-border data flows, and review standards may continue to evolve. Periodic adjustments to relevant rules may have a certain impact on companies' R&D planning and internationalization pace.