A representative from Huawei's pharmaceutical division stated that the company does not intend to develop medicinal drugs or create its own production chains. Speaking to the press at the XI China Conference on Pharmaceutical Innovation and Investment in Nanjing on October 24, he clarified that instead of this, Huawei aims to serve as the technological foundation for drug discovery using artificial intelligence.
The company is ready to provide drug manufacturers, research institutes, and AI-driven drug discovery firms with computing power, data storage, network solutions, toolkits, and data management mechanisms.
This division is one of Huawei's so-called legions—teams created by the company to focus on specific industries. It began appearing at pharmaceutical industry forums in late 2025.
Assessing AI Development in Drug Discovery
The leader described the process of AI-driven drug discovery as rapidly evolving but still holding high expectations. Some technologies, such as protein structure prediction and molecular modeling, are already in practical use, but the complete drug development pipeline has not yet been established. He noted that the next key question is not about model size, but about their ability to integrate into real research workflows and form a closed loop with laboratory data and researcher decisions.
The potential for AI to increase efficiency and help companies consistently release products will only become clear through actual production lines and resulting outcomes.
In partnership projects, clusters consisting of approximately one hundred accelerator cards designed for training, deploying, and optimizing drug discovery models have already begun to be implemented. Huawei has built such clusters with two medical companies and supports scientific institutions, including the Shanghai Maternal Medicine Institute and the Chanping Laboratory.
Advantages of Huawei's Offer
He listed three advantages offered by Huawei. Firstly, it is access to advanced computing and data storage capabilities that many companies lack. Secondly, it is the ability to cluster these computations. Training a model requires a minimum of 64 cards, and often hundreds or thousands, which must be combined into a cluster rather than used individually. He also pointed to Huawei's development—the SuperPoD design with 1024 cards, which he called the largest of its kind in the world.
Thirdly, there is the ecosystem: Huawei collaborates with DP Technology, StoneWise, and MoleculeMind to create joint solutions. Huawei is responsible for coordinating and adapting computing, storage, and networking, while partners handle the development of the models themselves.
This stance reflects Huawei's approach to other industries, where the company provides infrastructure, leaving applications to partners. In the field of drug discovery, this means betting on the demand for large training clusters as more companies transition from small experiments to industrial-scale models.
