Silexon AI: Harnessing Artificial Intelligence to Overcome Pharma's Eroom's Law

Recently, the AI-powered innovative biomedical R&D platform Nanjing Silexon Intelligent Technology Co., Ltd. (hereinafter referred to as "Silexon AI") announced the completion of its Series A financing of over RMB 100 million. The round was jointly led by Sida Capital and Sequoia China, with co-investment from Quanchuang Capital, and additional funding from existing shareholder Huiyu Capital. This marks the third round of financing for Silexon AI since its establishment in 2018, bringing its cumulative financing over the past three years to hundreds of millions of RMB.

The capital frenzy surrounding AI drug discovery companies reflects an urgent desire to enhance the efficiency of new drug research and development.

Despite decades of continuously increasing investment by pharmaceutical companies, the number of new drugs approved per $1 billion invested is halved every nine years, a phenomenon known as "Eroom's Law" (also referred to as "Reverse Moore's Law").“Eroom” is the reverse spelling of “Moore,” meaning the opposite of Moore’s Law. The latter states that, with prices held constant, the number of components that can be accommodated on an integrated circuit will approximately double every 18 to 24 months, and performance will likewise double.

On July 17, 2019, the journal *Trends in Pharmacological Sciences* published a review article titled “Artificial Intelligence in Clinical Trial Design” by the AI team at IBM Watson Health, which pointed out that AI can accelerate the success of clinical drug trials, thereby helping to overcome the “Eroom’s law” dilemma.

So, how has Silexon been leveraging AI to crack the "Eroom" dilemma? What exactly is its core appeal that consistently attracts investor interest? How does the company bridge the two equally complex and demanding fields of drug discovery and artificial intelligence to accelerate the new drug development process?

Silexon CEO Zeng Hainianstated that, currently in the AI drug discovery sector, the common challenges and issues facing the industry are data, algorithms, and validation.

▷ DataThe problem is how to obtain a large volume of high-quality, highly reproducible, and low inter-batch variability positive and negative sample data relevant to a specific problem;

▷ AlgorithmThe key question is to select which framework or algorithm can efficiently and accurately extract the latent features or patterns of a specific problem, learn underlying regularities for novel predictions, and reveal the corresponding mechanisms (the interpretability of model prediction results), which would be optimal;

▷ Validation..., the question is which test datasets to use and how to conduct experiments to rapidly validate the stability, accuracy, and efficiency of the algorithm model.

Delivering practical and feasible solutions to the aforementioned challenges directly corresponds to Silexon's core competitive advantages and business model.

Since its establishment, Silexon has continuously advanced collaborative development and technology implementation in the biopharmaceutical sector, driving the deep integration of its AI technology with the industry to empower new drug research and development.

To date,Silexon has signed & reached cooperation intentions with over 20 renowned domestic and international pharmaceutical companies/CROs/institutions.These partners include Junshengtai, a pioneer in innovative drug R&D for chronic diseases; Abogen, a leading domestic enterprise in nucleic acid drug development; renowned CROs Viva Biotech and BioDuro; as well as other listed pharmaceutical companies and multinational pharma giants... Each is a standout company in its respective field that maintains exceptionally high standards for its partners.

These collaborations have provided Silexon with high-quality data and validation results. Through this in-depth partnership, Silexon is continuously optimizing its models to assist partners in more efficiently identifying superior candidate molecules.

In September 2018, Silexon was established under the incubation of the Turing Institute for Artificial Intelligence. Founded in April 2018, the institute is committed to providing public technical services through industry-academia-research collaboration, gathering and cultivating AI talents with interdisciplinary capabilities, and promoting the translation and industrialization of scientific and technological achievements.

Since its inception, the Turing Institute of Artificial Intelligence has provided Silexon with comprehensive, multi-faceted support, encompassing talent, technology, office space, and policy guidance.

The founding team of Silexon is among the earliest teams both in China and internationally to apply deep learning to genomics research, with years of experience in applying AI to drug discovery. The team has published a total of over 70 papers in internationally recognized core journals and conferences, including *Nature*, *Nature* sister journals, and *Cell* sister journals.

All team members are outstanding talents from numerous top-tier domestic and international universities, including Tsinghua University. Silexon’s profound technical expertise and professional capabilities enable it to continuously attract top talent across interdisciplinary AI fields. The integration of these elite professionals forms the foundation for the Silexon team to develop powerful algorithm models.

Throughout the continuous development of new models, Silexon’s algorithm R&D team consistently ensures the timeliness of model updates. Zeng Hainian told VCBeat, “We essentially update and optimize our training data and existing models on a weekly basis to ensure that the AI models remain in a state of ‘lifelong learning’ and ‘continuous iterative optimization.’ Furthermore, unlike many technology platforms that operate purely from an algorithmic perspective, Silexon has prioritized model interpretability as a key objective right from the initial stages of model development.”

To improve the predictive accuracy of tasks, Silexon employs multiple models to conduct cross-validation on the same task.

Furthermore, the Silexon team possesses a robust background in the biopharmaceutical sector, and the team is currently...RNA, DNA, Protein Folding, Computational Chemistry, and Drug-Target InteractionsHas already conducted extensive cutting-edge research and accumulated substantial research experience, publishing dozens of highly cited and high-impact peer-reviewed articles.

In routine R&D operations, the team conducts project initiation in accordance with biopharmaceutical company standards. This ensures that when Silexon designs and develops AI-driven drug discovery models, it can thoroughly identify and address the pain points and technical bottlenecks in novel drug development, thereby enabling deep integration into the AI-enabled drug R&D workflow. The Project Initiation Committee evaluates proposals across multiple dimensions—AI, medicinal chemistry, biology, and commercialization—to ensure that all corporate initiatives are rigorously justified and thoroughly validated from the outset.

Throughout the project implementation, team members with interdisciplinary backgrounds mutually enhance and elevate each other through continuous exchange. While undertaking tasks aligned with their respective expertise, they achieve professional complementarity through collaborative research and development.

The interdisciplinary R&D team, characterized by robust capabilities, seamless collaboration, and complementary expertise, has already delivered remarkably impressive results.

Independently developed by SilexonAI4D™/AI4Pat™ can identify preclinical active molecules at approximately 1/3 to 1/10 of the time and cost of traditional drug discovery methods, significantly enhancing new drug R&D efficiency, shortening development timelines, and reducing costs.

Specifically, the AI4D™ drug discovery platform primarily empowers the preclinical R&D stages for Class 1 innovative drugs, undruggable targets, and drug repurposing. It covers a series of essential steps, including target development, hit discovery, lead screening, lead optimization, and PK/PD assessment, aiming to efficiently obtain “First-in-class” candidate molecules with a high success rate.

The AI-assisted platform Silexon®AI4Pat™ was developed primarily to facilitate the R&D of fast-follow drugs, enabling the efficient, rapid, and cost-effective discovery of preclinical candidate molecules, thereby empowering pharmaceutical companies to leapfrog from "fast-follow" to "best-in-class".

Certainly, beyond the robust capabilities of the technical R&D team itself, Silexon’s rapid growth and highly efficient operations—driven by strong execution—owe much to an outstanding helmsman who strategically guides the company from behind the scenes. Silexon CEO Zeng Hainian has over a decade of hands-on experience in the pharmaceutical and healthcare industry, boasting extensive industry expertise.

His educational background—a bachelor’s degree from the School of Life Sciences at Fudan University, alongside a master’s in Regulatory Science for Biological Sciences from Johns Hopkins University (JOHNS HOPKINS UNIVERSITY) and a dual master’s in Plant Biology from North Carolina State University (NORTH CAROLINA STATE UNIVERSITY)—has equipped him with a solid foundation in FDA regulatory affairs, life sciences, and chemistry. Frontline industry experience, including his roles in business development (BD), strategic planning, and R&D management at Sinopharm Group, as well as nearly a year of regulatory review work at the Shanghai Medical Products Administration, has enabled him to cultivate a deep industry network and accumulate profound professional expertise. Furthermore, his tenure at Ping An Capital, where he managed primary and secondary market investments in the innovative biopharmaceutical sector, allowed him to rapidly familiarize himself with cutting-edge biopharma innovations and gain a deeper understanding of financing and capital operations. This rich industry background empowers Zeng Hainian to implement corporate decision-making and operations from a more open and long-term perspective, while his broad professional network helps Silexon more easily attract and consolidate top-tier talent.

Frequent investor backing has seemingly spared Silexon from ever encountering cash flow shortages or "financial tightness." The real challenge, however, lies in how to allocate every penny wisely to maximize value and returns. Zeng Hainian told VCBeat that in the near future, Silexon will continue to prioritize platform development, continuously optimizing the performance of its existing model tools while relentlessly pushing boundaries to develop new "breakthrough" models.

Moving forward, Silexon aims to leverage its comprehensive AI-powered drug discovery and development platform to establish in-depth collaborations with a growing number of pharmaceutical companies, thereby reshaping the entire new drug R&D pipeline, increasing the success rate of novel drug development, reducing R&D costs and timelines, and ultimately enhancing drug accessibility to benefit society.