AI and small nucleic acid, two cutting-edge technological waves are deeply intersecting.

In the first half of 2026, there will be over 10 global transactions related to small nucleic acid testing, with a potential total transaction amount approaching billions of US dollars; At the same time, AI for Science is moving from the laboratory to the forefront of the industry. Driven by the joint efforts of two waves, the research and development of small nucleic acids has entered a new stage of development.

In the past, the development of small nucleic acid drugs was limited by bottlenecks such as molecular design and delivery efficiency; Nowadays, AI is pushing the boundaries of this field, from sequence design to complex combination optimization, from intrahepatic delivery to extrahepatic delivery, and the innovation space continues to expand.

During the first Global Conference on New Drugs in Major Countries (CPIC 2026), Tongyi interviewed Dr. Gao Xiangrui, the head of nucleic acid drug research and development at Jingtai Technology, to explore how AI can reconstruct the siRNA (small interfering RNA) research paradigm, how platform capabilities can truly be implemented in pipelines, and the future direction of industry competition.

01. Wave Intersection: AI+Small Nucleic Acid Embarks on a New Industrial Cycle

In June of this year, Alnylam signed a $2 billion partnership with AI pharmaceutical company Inceptive Nucleics. The layout of global small nucleic acid heavyweight players has instantly attracted industry attention.

In Gao Xiangrui's view, this is not only a commercial transaction, but also signifies a change in the direction of industry development - AI is gradually evolving from a research and development tool to a key capability for driving siRNA innovation.

In fact, Alnylam's layout is not an exception. In recent years, large pharmaceutical companies and technology leaders around the world have been increasing their deployment of AI pharmaceuticals.

DeepMind, Anthropic and other companies continue to lay out AI for Science, while MNCs such as Pfizer and Sanofi are also exploring the application of AI in drug development through self research, cooperation or mergers and acquisitions. Gao Xiangrui believes that as more participants enter, AI pharmaceuticals will gradually see value realization in the next one or two years.

At the same time, the research and development of small nucleic acids has also reached a new turning point. As GalNAc mediated intrahepatic delivery gradually matures, the industry is exploring more challenging directions such as extrahepatic delivery and dual target design, which also provides new application space for AI technology to enter small nucleic acid research and development.

Moreover, according to Gao Xiangrui's analysis, AI has a natural advantage in siRNA design in terms of molecular characteristics.

Large molecule drugs involve complex three-dimensional structures, protein flexibility changes, and other issues, while siRNA itself is a sequence type molecule. The Transformer architecture commonly used in current Large Language Modeling (LLM) is adept at handling serialized data. ”Gao Xiangrui explained.

However, for AI to truly enter the small nucleic acid research and development process, it is not just a matter of algorithm capability, but also needs to be combined with experimental systems.

Gao Xiangrui mentioned that there is currently a view in the industry that relying on high-throughput experimental technology for large-scale screening can also promote the development of small nucleic acid drugs, so AI is not necessary.

However, Gao Xiangrui pointed out that AI and high-throughput experiments are not a substitute relationship, but a synergistic relationship. In this process, AI can help researchers quickly explore larger molecular design spaces and provide clearer optimization directions for experiments; High throughput experiments can validate AI design results and continuously generate new experimental data, further promoting model optimization.

AI design has a specific directionality, AI designs first, and then uses high-throughput methods to detect, which is like adding wings to a tiger. ”Gao Xiangrui said.  

02. Technical restructuring: Creating a hardcore barrier for AI nucleic acid research and development

AI reshapes the paradigm of small nucleic acid research and development, not only by improving the efficiency of individual links, but also by whether it can run through the entire process from molecular design, experimental verification to candidate drug screening.

Around siRNA research and development, Jingtai Technology has created Kodexia, a dry wet closed-loop development platform based on generative AI and first principles ™， I hope to improve the efficiency of siRNA drug development through the combination of AI and experimental systems, and promote the transformation of computational design into pipeline results.

Taking AlphaFold as an example, Gao Xiangrui pointed out that the core of its breakthrough is not simply to increase the number of parameters, but to deeply understand the inherent mechanism of protein structure and embed domain knowledge into model design. The development of siRNA follows the same logic.

Although siRNA belongs to sequence molecules, drug development is not simply a sequence matching problem, but also involves multiple factors such as RNA thermodynamics, gene silencing mechanisms, chemical modifications, delivery efficiency, and safety. Kodexia ™  Integrating these underlying rules into model design, comprehensively optimizing multiple indicators such as activity, long-term effectiveness, off target risk, and drug efficacy, rather than simply pursuing performance improvement.

This is also an important difference between AI and traditional computing tools.

In the past, computing tools were mostly used to provide assistance at a certain stage of the R&D process, such as helping to find potential targets or predict some molecular features. But truly advancing a drug requires solving multiple complex problems simultaneously: finding effective molecules while ensuring their potential for further development. The value of AI lies in helping developers find better solutions in the vast design space.

Having algorithmic capabilities alone is far from enough to achieve sustained and stable R&D conversion. According to Gao Xiangrui's analysis, there is a natural knowledge gap between AI technology and drug development, and cross team collaboration is particularly difficult. Even industry leaders like Alnylam, who are deeply involved in the small nucleic acid field and have complete experience in drug development and commercialization, face the challenge of adapting to the local environment and need to seek external technical cooperation when laying out AI research and development.

In his view, AI pharmaceuticals are not simply applying AI tools to traditional R&D processes, but require a simultaneous understanding of algorithm logic and drug development laws. This is precisely where Jingtai's advantage lies.

Gao Xiangrui explained that Jingtai itself is a cross disciplinary team composed of AI experts and pharmacology and drug chemistry experts. After years of collaboration, both parties can better transform the professional knowledge in drug development into AI capabilities. At the same time, the company also places greater emphasis on the role of data in model optimization, driving AI capability iteration through continuous accumulation of data.

High quality self-developed data is the foundation for the continuous evolution of the model. Currently, many AI models rely on open-source public datasets for training, with public data highly concentrated on mature targets such as PCSK9 and HER2. The model is prone to good performance metrics in popular targets, but when targeting new targets with innovative potential, available data is scarce and the model's generalization ability is full of uncertainty.

We can't always rely on open source data because there is relatively little open source data and the diversity is also poor. Doing it and doing it may eventually become a benchmark, but it won't be very helpful for promoting the drug pipeline, "Gao Xiangrui admitted.

Self built dry wet closed loop is the core path to continuously produce high-quality in vitro and in vivo experimental data. The closed-loop system continuously produces real experimental results, including not only valid results but also a large amount of negative data. Gao Xiangrui stated that negative data is also an important foundation for AI model iteration, and only by continuously accumulating real experimental data can the model continuously optimize its predictive ability.

Currently, relying on Kodexia ™， Jingtai's IgA nephropathy siRNA pipeline obtained non-human primate (NHP) efficacy data within 7 months, with better activity and long-term efficacy than clinical reference molecules targeting the same target. It is expected to determine PCC (preclinical candidate compound) molecules within 9 months after project approval, significantly shortening the research and development cycle compared to the industry average of 12 to 18 months.

In addition, AI's powerful parallel computing capability has broken the efficiency constraints of traditional serial research and development. Traditional R&D is difficult to simultaneously undertake a large number of projects, while AI can parallel complete the design optimization of multiple targets and sets of molecular schemes, and the marginal cost will not increase significantly with the increase of projects.

Relying on its self-developed AI platform capabilities, Jingtai is practicing a multi-target parallel research and development model, deploying 6 new targets for the same disease field, and planning to obtain preclinical candidate compounds for all 6 targets within 12 months.  

03. Forward looking breakthrough: differential layout to seize the future increment of the industry

In recent years, small nucleic acid drugs have become a hot track in the field of innovative drugs. However, with the active layout of various enterprises, industry competition is changing. It is not easy to achieve true innovation and future growth.

Specifically, in the past few years, intrahepatic delivery technologies represented by GalNAc have driven the rapid commercialization of siRNA drugs, with a large number of companies focusing on mature targets such as metabolic diseases. However, as the number of participants increases and competition for popular targets intensifies, it has become increasingly difficult to establish long-term advantages through follow based development.

At the same time, top overseas enterprises have established a comprehensive patent system around RNA sequence design, chemical modification, and delivery technology after years of accumulation. For newcomers, optimizing only in mature directions not only faces fierce competition, but also easily encounters intellectual property barriers.

Gao Xiangrui believes that the truly valuable innovation increment in the field of small nucleic acids in the future will come more from previously difficult to break through new targets, new delivery methods, and new molecular design strategies.

In terms of innovative targets, dual target siRNA is also one of the directions he emphasized. Compared to traditional single target drugs, dual target drugs require simultaneous consideration of the synergistic effect of two siRNA molecules, as well as multiple factors such as linker (linking structure) and delivery, making them essentially more complex combinatorial optimization problems.

It is difficult to rely solely on expert experience to solve such problems, and AI is precisely good at handling complex combinatorial optimization. By continuously generating design schemes and learning experimental data, AI is expected to drive breakthroughs in the development of dual target siRNA. ”Gao Xiangrui pointed out.

In addition to target design, extrahepatic delivery is also a key technology that determines the future space of the small nucleic acid industry.

At present, the liver is still the most mature application scenario for siRNA drugs, but the delivery efficiency of extrahepatic tissues such as the central nervous system, lungs, kidneys, etc. still hinders the development of the industry.

In Gao Xiangrui's view, the breakthrough in extrahepatic delivery is not only about optimizing siRNA molecules themselves, but also involves the design of delivery targets. Due to the large exploration space of the delivery system itself, R&D personnel need to constantly try new design solutions and conduct screening and verification. The value of AI lies in helping R&D personnel generate more potential solutions, combined with high-throughput experiments for screening, thereby improving the efficiency of discovering effective solutions.

Following this breakthrough strategy, as of now, Kodexia from Jingtai ™  We have completed the transition from concept validation to pipeline assets, and have laid out 5 differentiated siRNA pipelines in cutting-edge directions such as metabolism, nephropathy, dual target, and extrahepatic delivery. Among them, more than half of the pipelines have completed in vivo efficacy evaluation, with excellent animal data performance, and the fastest pipelines have entered the PCC stage. The steady advancement of these pipeline assets validates the platform's research and development capabilities in multi-target, extrahepatic delivery, and complex disease scenarios.

In the past year or two, small nucleic acid has become a globally recognized next-generation innovative drug track with its unique therapeutic advantages. Based on this wave of industrial transformation, Jingtai Technology hopes to work together with global industry partners to explore the innovative potential of nucleic acid drugs.

On the commercialization path, Gao Xiangrui stated that Jingtai adheres to an open and cooperative attitude, and connects with domestic and foreign pharmaceutical companies through various models such as license out and co development to jointly promote the landing and transformation of siRNA technology with source innovation.

Innovation in nucleic acid testing is only one part of Jingtai's intelligent drug development layout. As a pioneer in AI pharmaceuticals, Jingtai is expanding its mature intelligent research and development capabilities to multiple drug modalities, continuously breaking through research and development bottlenecks, expanding the boundaries of potential drugs, and exploring more possibilities for innovative drugs.