Guotai Junan Securities: A panoramic view of global AI pharmaceuticals from the perspective of three types of leaders in data, computing power, and models

According to the Zhitong Finance APP, Guojin Securities released a research report stating that the breakthrough of AI new drugs is imminent, with a preference for those with rich pipelines and strong realization capabilities. As the singularity of the AI pharmaceutical industry approaches, the first significant milestone will inevitably be the approval and market launch of the first AI-driven drug developed by humans; in other words, regardless of whether the model or data is superior, the first validated blockbuster drug will be the focus. Additionally, there are many entrants from pharmaceutical and cross-industry companies, with preferred barriers thickening over time. This is because AI drug development itself is a brand new track of technological crossover, and the future first breakthrough could be an AI pharmaceutical company, a traditional leading generic innovator deeply engaged in the AI field, or a new technology company from a non-pharmaceutical sector.

The main points of Guojin Securities are as follows:

Timing

The application of AI from concept to reality, multi-omics development reduces costs and increases efficiency by 1000 times, and the first AI blockbuster drug is about to emerge. (1) Time has come: Human thoughts on AI began with the Turing Test in 1950. With DeepMind's AlphaGo defeating the world Go champion in 2016, and Hassabis and Jumper winning the Nobel Prize in Chemistry in 2024 for the successful prediction of protein structures by AlphaFold2, the era of AI drug development has already begun. (2) A qualitative change is imminent: According to the "Big Idea 2025" report released by Ark Invest, which focuses on investing in disruptive technologies, a disruptive quantitative change is occurring in the biopharmaceutical field. Multi-omics AI applications will bring a 1000-fold reduction in costs and increased efficiency in the pharmaceutical field. The era of super Moore's Law for innovative drugs has begun. (3) Problems have been solved: AI models, the black box has been broken. The reproducibility and verifiability of any scientific research, including innovative drug development, whether in vitro or in human clinical trials, are extremely important. The EU's "Artificial Intelligence Act" will eliminate AI drug discovery systems that rely on black box models and lack interpretability. Leading AI pharmaceutical companies have already moved beyond concept validation. InSilico Medicine's TNIK, ENPP1, and PHD have successfully advanced through three replicative paths, with the AI development process detailed in Nature magazine.

Essence

Cloud access to computing power, limitations of data quantity and quality have all been broken, and models are thickening barriers over time, leading to a rebound and acceleration in the return on innovative R&D. (1) Computing power is the premise for the operation of AI applications. Global tech giants, including Amazon, Google, Microsoft, and Alibaba, have ample cloud computing power available for pharmaceutical companies to choose from. Meanwhile, NVIDIA will launch NVIDIASpectrum-XGS Ethernet on August 22, 2025, a cross-scale technology for combining distributed data centers into a unified gigabit-level AI super factory. This move will further enhance the global computing power level. (2) Data: From DL (Deep Learning) to FL (Federated Learning), MNCs (multinational corporations) are joining forces, breaking through the data cocoon. On one hand, since Google introduced FL in 2016, this technology has received widespread attention. This algorithm aims to build better models and is driven by multiple aspects, including the benefits of decentralized learning processes across a group of devices (cross-device FL), accessing widely distributed knowledge (cross-island FL), and protecting the privacy of local data. On the other hand, companies like Apheris and the UK government are integrating cross-industry collaboration to break through the data limitations of AI-driven drug discovery According to a press release from the UK government website on June 9, 2025, the UK "OpenBind" alliance announced that it will utilize breakthrough experimental technologies to generate the world's largest dataset of drug and protein (components of the human body) interactions. (3) The model is the key to success. For generative AI pharmaceutical companies, building leading model barriers that grow over time from hypothesis to realization is crucial. When computing power and data do not pose significant constraints, the efficiency of model development iteration and the advantage of accumulated training experience become vital. The positive feedback loop of data and model construction in the AI research and development process with large pharmaceutical companies will create the moat for currently leading enterprises.

Industry Changes

Tech giants are entering the field, and industry chain companies are accelerating their configurations; the top ten pharmaceutical giants are all involved in AI layout. (1) Tech giants: NVIDIA is promoting BioNeMo and investing in AI pharmaceutical companies. Google has acquired DeepMind and split Isophormic Labs, which publicly stated in July that clinical trials are "very close." Tech giants are pouring into AI pharmaceuticals. (2) Industry chain: Hongbo Pharmaceuticals is promoting the DiOrion platform, deeply empowering IND (Investigational New Drug) compliance submissions, while InSilico Medicine is fully integrated from hypothesis, molecules, clinical trials to release. (3) Pharmaceutical giants: Leading companies such as Merck, Pfizer, Eli Lilly, and BMS have invested hundreds of billions of dollars in AI pharmaceutical-related companies. According to statistics from the Pharmaceutical Cube, the top 20 projects related to AIDD (AI Drug Development) in terms of initial and total transaction amounts have seen significant transactions concentrated in the past five years, with a total exceeding $50 billion.

Risk Warning

Risks related to international exchange rates, fluctuations in domestic and foreign policies, volatility in investment and financing cycles, and risks of mergers and acquisitions not meeting expectations