Nomura - China Healthcare: AIDD... Nomura - China Healthcare: AIDD expert call… · @2030FY_X · TwKit

@2030FY_X

Nomura - China Healthcare: AIDD... Nomura - China Healthcare: AIDD expert call notes - Current AI applications concentrated in preclinical stage, data assets and closed-loop capabilities form the core watershed Nomura hosted two expert calls on AI-driven drug discovery (AIDD), inviting an AI-enabled CRO serving external pipelines (Expert A) and an AI-native biotech with its own R&D pipeline (Expert B) to participate in the discussion. The two experts reached a consensus that "AI is currently mainly applied in the preclinical stage" and "data moats are higher than model moats," but showed significant divergence on the sustainability of business models. Core conclusions are as follows: AI currently only penetrates the preclinical stage, which accounts for a small proportion of the entire drug R&D life cycle (such as target optimization based on binding affinity prediction), and only small molecule drug candidates have truly entered the clinical stage. Because the models and experimental data of various companies are highly opaque, it is difficult for external investors to horizontally compare technological superiority. The number of academic publications, the scale of interdisciplinary scientists, the frequency of partnership deals, and the advancement status of proprietary pipelines (for example, Insilico Medicine's Rentosertib entering Phase III was specifically highlighted) are among the few viable proxies to verify their true strength. The popularization of open-source models will weaken the scarcity of generic data, while proprietary experimental data (wet lab data) derived from proprietary pipelines is becoming irreplaceable. Therefore, Expert B is strongly bullish on the long-term prospects of AI-native companies, believing they possess recurring order resilience brought by the "dry lab + wet lab" closed loop, whereas AI-enabled companies are closer to project-based outsourcing, with uncertainty in recurring revenue. Furthermore, Expert B pointed out that AI labs recently crossing over into the AIDD space generally lack life science accumulation and wet lab capacity, making it difficult to shake the existing landscape in the short term. Nomura believes this note provides investors with a key micro perspective to separate "true moats" from "concept hype" in the AIDD track. Core point breakdown 1. Technology application boundaries: Preclinical is the main battlefield, breakthroughs in biochemical understanding are the next inflection point Both experts unanimously judge that the value of AI in pharmaceuticals is currently highly concentrated in the preclinical stage, a segment that accounts for a limited proportion of overall time and capital investment. The main application scenarios are target optimization and molecule generation based on binding affinity prediction, and truly AI-designed small molecule candidates have only just entered the clinic. To achieve disruptive breakthroughs, AI must achieve an underlying paradigm upgrade in understanding complex biochemical pathways and bioinformatics, rather than merely relying on the fitting of existing data patterns. This means that over the next 2 to 3 years, AIDD's commercialization contribution will remain confined to improving early-stage R&D efficiency, rather than shortening the overall cycle. 2. Competitive moat reconstruction: Proprietary wet lab data replaces generic models as a scarce resource As open-source AI models lower the barriers to entry for generic computing power and foundational algorithms, the moats of companies possessing purely "dry lab" modeling capabilities are narrowing. Expert B emphasized that AI-native companies with proprietary pipelines that continuously generate high-quality wet lab feedback data can constantly self-iterate their data closed loop, and such proprietary data assets cannot be replicated by external competitors. Although Expert A partially agreed, they also noted that AI-enabled companies still hold service value in the early screening stage. From a quantitative perspective, Nomura believes that when evaluating an AIDD company, higher weight should be given to "the number and stage of proprietary pipelines" and "capital expenditure on wet lab automation equipment," which are better predictors of long-term gross margin stability than simply counting partnership deals. 3. Business model divergence: AI-native companies vs. AI-enabled CROs Expert B's core bull logic is that AI-native companies naturally form recurring orders and data backflow through the advancement of internal pipelines, offering higher revenue visibility; whereas AI-enabled companies are closer to the traditional CRO pay-per-project model, facing customer demand fluctuations and price comparison pressure. Expert A acknowledged this risk but believes the two are complementary at this stage—the former excels in early dry lab exploration, while the latter possesses wet lab validation capacity. Expert B further poured cold water on this, arguing that the results of traditional pharma companies building their own AI platforms are far inferior to those of specialized early-stage AI-enabled companies, and newly entered pure AI labs will struggle to cross the "valley of death" from computation to validation due to a lack of wet lab capabilities. 4. Investment implications and observation metrics In the current environment of valuation divergence, Nomura advises investors to avoid targets that rely solely on public paper narratives, and instead track the following hard metrics: first, pipeline asset progress, especially the number of AI-originated molecules entering clinical Phase II and above; second, wet lab automation equipment coverage and capacity utilization, which directly determine the speed of the data flywheel; third, the repeat signing rate of partnership deals, used to verify customer stickiness. For AI-native companies, the market should grant a valuation premium corresponding to their closed-loop capabilities; for AI-enabled companies, it is necessary to rigorously scrutinize the depth of their penetration into front-end R&D, rather than just looking at short-term order flow. Summary The market has previously overly focused on the superiority of model parameters, but industry experts clearly point out that data closed loops and wet lab capabilities are the key to divergence over the next three years. Nomura believes that with the democratization of open-source tools, AIDD competition will shift from an "algorithm race" to a race of "data quality and experimental closed-loop efficiency." In the current valuation repair window for the healthcare sector, it is recommended to overweight AI-native enterprises with proprietary pipelines and scaled wet labs, maintain caution on pure outsourcing AI-enabled vendors, and be wary of their profit margins coming under pressure in price competition due to a lack of data moats.

曝光 198 · 评论 0 · 点赞 2 · 书签 1