MetaNovas Secures Two Rounds of Series A+ and A++ Funding, Accelerating New Material Development with an "Army" of Agents | Exclusive Report by 36Kr
36Kr learned that MetaNovas recently completed two rounds of Series A+ and Series A++ financing, jointly invested by consumer health industry capitals such as Fuhua Capital, Hillhouse Capital, and Kangaroo Mother Group. Previously, the Series A financing was jointly led by Hillhouse Ventures and Baoding Ventures, with Ruoyuchen as a follow - on investor.
Today, the upsurge triggered by AI For Science continues. AI is accelerating the transformation of the development paradigms of industries such as new drugs and new materials. However, the market is gradually returning to rationality: although the difficulty of generating and optimizing molecular structures through AI has been significantly reduced, the challenges at both ends have not undergone qualitative changes due to technological evolution.
Under the traditional technical path, it takes about 3 - 5 years to develop a consumer - type new material, from the initial product selection, mid - term large - scale production, to the back - end registration and commercialization. To improve the efficiency of the entire chain, MetaNovas has built a system - level operating platform centered on Agentic AI (Intelligent Agent Artificial Intelligence) to make multi - objective decisions in a highly uncertain R & D environment, taking into account the performance, process requirements, and regulatory constraints of new material molecules, and reducing the cost of commercialization from the source.
Wang Meijie, the co - founder and CEO of MetaNovas, told 36Kr that driven by the "AI Intelligent R & D Organization", the fastest - developed bioactive raw material completed the process from concept proposal to human efficacy testing within 12 months. Thanks to the application of AI agents in the entire chain, MetaNovas can maintain a lean and efficient team and support the manpower required for the rapidly increasing new material pipelines.
Currently, AI algorithm experts, biological teams, and transformation teams each account for one - third of the MetaNovas team. The founding team members all have a composite background in AI, biomedicine, and computational materials. Wang Meijie once worked at NVIDIA's Silicon Valley headquarters, developing artificial intelligence infrastructure for biological computing; Yu Lun, the Chief Technology Officer, is a doctor in nuclear science and engineering and AI from MIT and once served as the Chief Data Scientist at UnitedHealth Group in the United States.
It is reported that MetaNovas has self - developed a large molecular language generation model as the underlying generation engine, which can represent polypeptides, polymers, small molecules, etc. across modalities, "covering a chemical space of more than 10^60 with a molecular generation efficiency of over 95%". At the same time, for the physical and chemical properties (such as thermal stability, odor, ultraviolet absorbance, etc.) that must be considered for material implementation, it has developed a performance prediction model to provide a basis for molecular screening.
The key to improving model accuracy is the accumulation of high - quality data and an active learning system that automatically iterates based on experimental data. Yu Lun introduced that the training data mainly includes three categories: literature and patent data; laboratory data authorized through cooperation with academic institutions; and high - throughput wet experiment data generated by the internal experimental platform. Among them, the self - owned experimental platform has not only accumulated successful verification data but also precipitated "failed" negative sample data. These scarce internal feedbacks make the AI system more accurate in the iteration.
Image source: MetaNovas
In order to enable the AI system to have the thinking and capabilities of an R & D team, MetaNovas has developed AI agents for the entire process of new material development, including literature mining, molecular generation, performance prediction, experimental planning, market, and commercialization.
"New material development involves teams with different backgrounds, including biologists, medicinal chemists, formulators, marketers, etc. The core of developing agents lies in building a workflow that can produce effective knowledge more efficiently, abstracting the dynamic collaboration mechanism developed through long - term cooperation among human teams into core steps and key quality control (QC) nodes that agents can directly execute. This depends on the know - how accumulated by the previous teams in each link." Yu Lun explained.
In the market insight stage, the Agent system will collect real - world data (ingredients, formulations, sales volume, etc.) from consumer product channels for forward - looking analysis. Before biologists design experiments, AI has already excluded overly competitive tracks based on market trends, guiding R & D towards more differentiated and market - potential directions, avoiding the sunk cost of "developing products that the market doesn't need".
Empowered by the Agent system, the first - time success rate of the molecules recommended by the MetaNovas platform exceeds 60%, significantly reducing the trial - and - error cost and the number of iterations. The Senoreversing peptide developed by it completed experimental verification after only testing 42 peptide molecules and going through 2 rounds of iterations, and this molecule has also attracted the attention of brand owners such as Unilever. In addition, the new antibacterial and anti - inflammatory molecule AMP33 designed by AI has obtained the medical device master file record.
As an AI - native new material development platform, MetaNovas is expanding its pipeline development directions, including bioactive ingredients, medical materials, functional polymers, photochemical ingredients, odor and flavor ingredients, etc. In terms of process scale - up and production, it mainly cooperates deeply with CDMOs and establishes joint ventures for targeted production transformation. Commercially, it mainly conducts business through models such as joint development with brand owners and material supply.
Driven by Agentic AI, materials science is bidding farewell to the long and expensive "blind screening era". When AI is no longer just a simple generation tool but evolves into an "intelligent R & D organization" that never gets tired, can cross disciplinary gaps, and understands commercial trade - offs, a new industrial era of new material R & D is coming.