In the past, new drug development was defined by the "Double Ten Law": an average R&D cycle of approximately 10 years, an investment of 1 billion US dollars, and a success rate of only 10%. As new drugs targeting easily druggable targets have been successively approved, the actual R&D cycle for drugs has been lengthening. According to Nature, it now takes an average of 10 to 15 years and costs about 2.6 billion US dollars to bring a new drug from R&D to market approval, with only a tiny fraction of candidate drugs ultimately succeeding in launching.
Faced with this harsh reality, pharmaceutical companies regard AI as a powerful tool to reshape the industry landscape. In recent years, players across the board-from MNCs and Biotechs to AI-native enterprises and even internet giants-have entered the AI-driven drug discovery arena.
The recent collaboration between Eli Lilly and NVIDIA is a prime example of the "hundred boats vying for the lead" dynamic in this field. The supercomputer developed through their partnership is powered by over 1,000 B300 GPUs, making it the world's first NVIDIA DGX SuperPOD to adopt the DGX B300 system. Beyond drug discovery, Eli Lilly will also use this supercomputer to shorten drug development cycles and expand the scope and complexity of its drug discovery efforts.
Beyond Eli Lilly, numerous other MNCs have laid out their strategies in AI-driven drug discovery. In the first half of this year alone, major collaboration deals in this sector reached nearly 10 billion US dollars, involving partnerships between multiple MNCs and AI platform companies.
In March this year, Novo Nordisk entered a collaboration with AI-driven drug discovery firm Gensaic. Novo Nordisk will leverage Gensaic's novel protein design technology to develop innovative drugs for metabolic and cardiovascular diseases. Gensaic owns a unique proprietary technology platform called FORGE, which combines unbiased protein evolution with machine-guided design. This platform identifies new mechanisms of action for tissue-specific intracellular delivery, enabling the development of drugs targeting novel therapeutic targets.
In September, Merck signed an agreement with AI-driven drug discovery company Variational AI. Under the agreement, Merck will use Variational AI's AI drug discovery platform to design and optimize small-molecule drug candidates for two undisclosed targets.
Additionally, Pfizer has integrated AI into drug discovery and clinical trials to optimize patient population analysis and the utilization of real-world data. AstraZeneca has partnered with Immunai, utilizing the latter's AI-powered immune system models to enhance the efficiency of anti-tumor drug trials.
Beyond multinational pharmaceutical giants, several internet and tech giants have also crossed into the AI-driven drug discovery space. Tencent launched "Yunshen Zhiyao," an AI-powered drug R&D platform; Huawei released the "Huawei Cloud Pangu" large model for drug molecules; Alibaba has built out AI-driven drug R&D and big data platforms; and ByteDance established Protenix, an AI drug R&D team-all demonstrating their commitment to advancing in this field.
These developments all indicate that starting this year, the "arms race" in AI-driven drug discovery has entered a new phase. Currently, the industry has already achieved several breakthroughs in this domain.












