According to a research report from Guotai Haitong (ASX: 601211), the key difference between this round of AIDD and the previous one lies in AI upgrading from a search efficiency tool to a system capable of executing a complete decision-making framework and designing molecules from scratch, which is expected to transform the underlying logic of drug design.
Since 2026, foundational large model developers have increased their investments in the healthcare sector, and pharmaceutical company collaboration models have shifted from project-based arrangements to enterprise-level deep co-building such as model subscriptions and joint laboratories. The industry has entered a critical window for clinical validation and order fulfillment.
The main views of Guotai Haitong are as follows: AI drug discovery has reached an industry inflection point, with large models entering the space to reshape the sector paradigm. The rapid development of large language models has accelerated scientific research discoveries including in the biomedical field.
The firm believes that the key difference between this round of AIDD and the previous one lies in AI upgrading from a search efficiency tool to a system capable of executing a complete decision-making framework and designing molecules from scratch, which is expected to change the underlying logic of drug design. Since 2026, foundational large model developers have increased their investments in the healthcare sector, and pharmaceutical company collaboration models have shifted from project-based arrangements to enterprise-level deep co-building such as model subscriptions and joint laboratories. The industry has entered a critical window for clinical validation and order fulfillment.
AIDD is expected to improve the druggability of candidate molecules and expand the space of biological hypotheses. The core pain point in pharmaceutical R&D lies in clinical translation. The success rate of corrected drug-indication projects has remained at a low level of approximately 8%-9% in recent years, with about 86% of clinical sunk costs coming from projects terminated at Phase II and beyond.
The firm believes that insufficient efficacy and safety issues largely stem from early-stage decisions regarding targets and candidate molecules. AIDD is expected to break through from two ends: on one hand, by simultaneously optimizing multiple molecular properties and improving druggability; on the other hand, by broadening the selection space for targets, epitopes, and molecular formats, expanding the number of projects entering the R&D funnel. Consequently, there is dual incremental market space: first, improved clinical success rates are expected to restructure the R&D cost model, with saved R&D expenses able to be redirected to more projects; second, the number of feasible projects itself expands.
A computational-experimental closed loop is forming, and wet lab services stand to benefit with certainty. The firm believes that small molecules have been relatively thoroughly developed on existing druggable targets, with new projects increasingly relying on breakthroughs at the target and mechanism level. AI-driven de novo design, meanwhile, allows large molecules to continue generating new projects on known targets by leveraging epitope and molecular format expansion, and the experimental label data produced by these projects (including both successful and failed samples) constitutes the most core material for current large model training and iterative optimization.
As a result, demand bottlenecks will transmit level by level down the chain of "AI model generates designs → DNA synthesis → protein production → experimental validation → preclinical testing." AI wet lab orders are gradually being fulfilled, and capability barriers continue to deepen. Wet lab data with real labels represents the most certain demand currently, with DNA synthesis and protein production becoming the most direct bottleneck nodes. GenScript (integrated Gene-to-Protein delivery) and Twist (large-scale sequence synthesis) are the first to benefit.
The firm believes that during the model training phase, supplier competition focuses on delivery throughput, turnaround speed, and data quality, with automation equipment becoming the primary constraint on current capacity expansion. In the medium to long term, the competitive focus will shift toward the preparation and characterization of complex targets and innovative modalities, as well as full-chain validation coverage capabilities. Additionally, the current pharmaceutical process has not yet been comprehensively restructured, and demand across various segments of the industry chain will be released sequentially, with downstream segments such as efficacy testing, safety evaluation, and CMC expected to absorb incremental demand as pipelines advance.
Risk warnings: risks of technology iteration falling short of expectations, risks of order and capacity fulfillment falling short of expectations, risks of intensifying industry competition, and risks related to data and supply chain security.