MaKe | MaHui Member DeepWise Raises Over RMB 700 Million in Funding

DeepWise, a MaHui member company, announced the completion of a new funding round exceeding RMB 700 million. The proceeds will primarily be used to continue attracting top industry talent, deepen its full-stack capabilities spanning from foundational algorithmic innovation to industrial software and sector-specific solutions, and accelerate the commercial deployment and market expansion of its technologies in biopharmaceuticals, new energy, chemicals, semiconductors, and other verticals. Source Code Capital was a co-lead investor in DeepWise's Series B round.

DeepWise is dedicated to applying the "AI for Science" paradigm to build next-generation multiscale simulation platforms and industrial infrastructure for the most foundational areas of human civilization — biopharmaceuticals, energy, materials, and information science and engineering. In recent years, DeepWise has entered a phase of accelerated scaling and commercialization, establishing itself as a leading force in the AI for Science sector. The company has developed a suite of computational engines and pretrained models addressing core scientific challenges across microscale and mesoscale simulation, molecular property prediction, and experimental characterization optimization. Building on these foundations, DeepWise has created next-generation AI and simulation-driven R&D solutions for scientific research, drug discovery, and battery development, while rapidly constructing an open ecosystem and industrial collaboration framework around AI for Science.

Weijie Sun, founder and CEO of DeepWise, noted that the world is fundamentally composed of atoms, and all human creation in the material world ultimately comes down to manipulating and combining just over 100 types of atoms. In the AI for Science era, he said, the accumulated experience and principles governing these manipulations can be effectively encoded into models and software. DeepWise aims to leverage this new paradigm to equip researchers with the hands, eyes, and brains needed to perceive and operate in the microscale and mesoscale worlds — transforming drug and battery R&D from large-scale experimental trial-and-error to a "computational design, experimental validation" model. The company seeks to provide effective software tools and solutions for numerous industrial sectors, generating new momentum for scientific progress, industrial upgrading, and China's strength in science and technology.

Source Code Capital maintains its commitment to investing in technology-driven innovation and the creation of enduring, real value. The firm continues to support fields tied to foundational scientific research and believes that cross-disciplinary convergence will yield significant innovation opportunities. It looks forward to the scientific transformation led by DeepWise and the AI for Science paradigm, and the profound impact this will have on industry.

DeepWise Builds Full-Stack AI4S Capabilities

DeepWise is a pioneer of the AI for Science research paradigm. Its original "multiscale modeling + machine learning + high-performance computing" approach has achieved a breakthrough unification of accuracy and efficiency in multiscale molecular simulation. With its next-generation molecular simulation technology DeePMD, core members of the DeepWise team received the 2020 ACM Gordon Bell Prize, the highest honor in high-performance computing. This work was also selected as one of China's Top Ten Scientific and Technological Advances of 2020 by the Chinese Academy of Sciences and Chinese Academy of Engineering, and one of the World's Top Ten AI Scientific and Technological Advances of 2020. DeepWise has been recognized as a national high-tech enterprise and a national "Little Giant" specialized and sophisticated enterprise.

Since its founding, DeepWise has never ceased innovating and delivering results. Over the past year, DeepWise and its collaborators have released a series of major technical achievements, iterating on AI for Science infrastructure and bringing the field into the pretrained model era: the Uni-Mol molecular representation large model won first place in the OGB-LSC quantum chemistry property prediction task and provides novel solutions for pharmaceutical, catalysis, and OLED applications; the Uni-RNA large model significantly enhanced capabilities across nearly all RNA-related R&D scenarios including mRNA vaccine design, RNA structure prediction, ASO development, siRNA therapeutic innovation, small-molecule drug discovery targeting RNA, and aptamer R&D; and the DPA pretrained model substantially improved the transferability and generalization of deep potential models, demonstrating leading performance and advantages in applications such as high-performance alloys, semiconductors, and battery materials design.

A Three-Dimensional Product Matrix Based on the "Four Beams, N Pillars" Framework

Advancing AI for Science and moving toward a "platform research" model requires addressing common challenges across different scientific domains and close collaboration between industry, academia, and research. At the 2023 Science for Intelligence Summit, Weinan E proposed building shared AI for Science infrastructure, summarized as the "Four Beams, N Pillars" architecture. The "Four Beams" encompass: algorithm models and software driven by fundamental principles and data; high-precision, high-efficiency experimental characterization methods; databases and knowledge bases as alternatives to traditional literature; and highly integrated computing platforms. The "N Pillars" refer to industrial design and simulation automation platforms for practical applications in drug discovery, batteries, chemicals, and other industries, all built upon the foundation of the "Four Beams."

Based on this platform architecture and years of accumulated foundational technical capabilities, DeepWise has developed industrial design and simulation infrastructure including the Bohrium® scientific research cloud platform, Hermite® drug computational design platform, RiDYMO® difficult-to-drug target R&D platform, and Piloteye™ battery design automation platform, along with accompanying industry solutions. DeepWise's strong capabilities in both research and industrial implementation are increasingly evident. The company has established partnerships with multiple leading customers in pharmaceuticals, materials, and new energy, building a complete ecosystem from foundational innovation to industrial deployment through the empowerment of its industrial design platforms.

In the research domain, Bohrium® serves as the gateway to AI for Science-era research, integrating open knowledge sharing, convenient simulation computing, and efficient collaborative development. It has already empowered hundreds of research groups from universities and research institutes. Through user-friendly cloud-based Notebooks, AI4S computational tools deeply optimized for cloud and HPC hardware, a rich variety of scientific computing apps, and abundant computing resources, Bohrium® significantly expands the boundaries of research productivity. By enhancing researchers' innovation efficiency in learning, application, and discovery, it has cultivated a generation of developers for the AI for Science era.

In biopharmaceuticals, DeepWise has partnered with numerous customers to more broadly integrate physics-based modeling + AI computational paradigms into preclinical drug R&D. Major listed pharmaceutical companies, leading innovative biotechs, and top domestic and international CROs — including Hansoh Pharmaceutical, BioDuro, Jacobio Pharma, Hongbo Pharmaceutical, Nhwa Pharmaceutical, InventisBio, and Shanghai Hengrui Medicine — have adopted DeepWise's solutions to improve R&D investment efficiency and decision quality.

In battery R&D, industry leaders including CATL and Do-Fluoride have noted that Piloteye™, built on strong engineering capabilities combined with innovative algorithms and deep industry know-how, pushes the boundaries of R&D innovation and narrows the gap between computation and experimentation. They also highlighted that Piloteye™'s operational convenience and user-friendly design effectively lower the barrier for R&D personnel to use simulation software, allowing precious energy to be focused on the industry's critical challenges.

From Academia to Industry: DeepWise Advocates for an Open AI4S Ecosystem

In 2022, DeepWise led the publication of the world's first industry-oriented AI for Science Global Development Outlook. An updated 2023 edition was released at the recent 2023 Science for Intelligence Summit, authorized for distribution by Xinhua Net. The Outlook, grounded in the "Four Beams, N Pillars" AI4S platform architecture, calls on researchers and practitioners across all sectors to break down barriers, build consensus, create connections, and form iterative collaborations — jointly constructing a comprehensive AI for Science industry-academia-research ecosystem and embracing a future of mutual benefit.

We are seeing more and more institutions and enterprises begin to engage with and participate in AI for Science, embracing the new research paradigm to capture the enormous gains in innovation efficiency and economic benefit it offers, securing an early advantage in future competition. The future of AI for Science is promising.