中文
BioGeometry, an AI-native biopharmaceutical company, recently announced the completion of its Series A+ financing, bringing total funding across its Series A rounds to more than US$100 million. The round was led by China Life Capital, with additional investment from existing backer Shanghai Biomedical Innovation and Translation Fund. Other participants included CDB Technology Venture Capital, Real Power Capital, Guofang Ventures and leading biopharmaceutical industry investors, while existing investor Fortune Capital also participated. Index Capital continued to serve as the exclusive financial adviser. The proceeds will primarily support the continued development of GeoFlow, at the molecular scale, and GeoCell, at the cellular scale, the company’s two biological world models; the expansion of automated high-throughput laboratories; and the advancement of its proprietary drug pipeline toward commercialization.
Connecting Molecules and Cells: Building Multiscale Biological World Models
Life sciences R&D depends on understanding relationships across biological scales: how molecular structures and interactions influence cellular function, and how cellular responses to interventions help elucidate molecular mechanisms of action. BioGeometry is developing a system of biological world models that connects molecules and cells. GeoFlow is an all-atom biomolecular world model that continues to integrate structure prediction, de novo design and multistep reasoning. GeoCell models cellular states and their responses to different perturbations. Together, they support cross-scale research spanning molecular design and the prediction of biological function.
GeoFlow: From Structure Prediction and Molecular Design to Complex Molecular Systems
GeoFlow has achieved three generational advances over two years:
V1 — June 2024. GeoFlow V1 represents three-dimensional molecular structures at atomic resolution, with performance comparable to AlphaFold 3 in protein–protein complex structure prediction.
V2 — April 2025. GeoFlow V2 was the world’s first model to unify structure prediction and de novo design. It became China’s first AI foundation model to demonstrate de novo antibody design, breaking the dominance of overseas closed-source models.
V3 — October 2025. GeoFlow V3 introduced, for the first time in the series, a multistep reasoning loop integrating generation, evaluation and optimization. Inspired by antibody affinity maturation, it autonomously refines molecular designs through in silico evolution. Across more than 20 targets, it achieved an average hit rate of nearly 20%, almost 100 times that of previous-generation AI methods, and reduced lead discovery time to less than three weeks.
The forthcoming GeoFlow V4 scales both model parameter count and training data in line with AI scaling laws. It delivers substantially higher accuracy than V3 on core tasks including structure prediction, interaction modeling and de novo design, providing a foundation for the design of more complex molecular systems.
GeoCell: From Molecular Design to Disease Biology
Molecular design is the starting point of drug development. A drug’s success depends on whether its target drives disease and which patients are likely to benefit. These questions require an understanding of disease biology beyond the properties of the molecule itself. GeoCell, BioGeometry’s proprietary AI virtual cell model, uses generative AI to model cellular transcriptomic responses to genetic and pharmacological perturbations, individually and in combination. It predicts changes in cellular states under previously unseen perturbation conditions and computationally reconstructs causal chains linking disease development and responses to intervention.
GeoCell thus supports three stages of R&D that depend most heavily on empirical judgment: target discovery, identifying targets that drive disease phenotypes; indication selection, determining the diseases in which a molecule is most likely to be effective; and patient selection, predicting which populations will respond to an intervention.
GeoFlow addresses which molecules to design, while GeoCell addresses target selection, disease indications and the patient populations most likely to benefit. Together, they form BioGeometry’s molecular and cellular world-model framework, deepening its understanding of disease biology as data and experimental validation results accumulate.
Recursive Self-Improvement of Biological World Models Through Computational–Experimental Feedback
AI development is moving from human-led training toward AI systems training other AI systems, a paradigm known as AI for AI. Recursive self-improvement (RSI) is regarded as a key pathway toward superintelligence. BioGeometry is applying this paradigm to life sciences, with each iteration of self-improvement validated through wet-lab experiments.
BioGeometry uses AI agents to connect its models with automated laboratories in an RSI framework integrating computation and wet-lab experimentation. The models formulate design hypotheses; AI agents automatically generate experimental protocols and orchestrate workflows; high-throughput platforms perform protein expression, purification and functional validation; and experimental data are fed back in real time to support model updates. These steps form a continuous design–experiment–learn cycle.
AI agents prioritize hypotheses most likely to reveal gaps in the models’ predictive capabilities. The models use experimental feedback to correct biases and improve molecular design. Model performance improves through iterative use, while the agents refine their ability to execute experimental workflows. This self-reinforcing AI-for-AI cycle in biology continuously generates proprietary experimental data that cannot be replicated by others.
The financing will support the construction of automated high-throughput laboratories, increasing experimental throughput and shortening feedback cycles to accelerate recursive self-improvement.
Translating Platform Capabilities into Commercial Value in Biopharmaceuticals and Synthetic Biology
BioGeometry has established a portfolio of dozens of proprietary drug development and synthetic biology product programs.
In biopharmaceuticals, several bispecific antibody programs are expected to reach the investigational new drug (IND) stage in 2027.
In synthetic biology, more than five product programs, including alpha-ketoglutarate (AKG) and natural borneol, have completed pilot-scale development.
BioGeometry reports the world’s first biosynthetic production of natural borneol, at a cost 80% lower than plant extraction. Commercial-scale production is expected to begin shortly, with substantial revenue anticipated in 2027.
The models advance the pipeline, and pipeline development informs further model improvement. As GeoFlow and GeoCell continue to improve, the RSI feedback loop develops and proprietary programs advance, BioGeometry is approaching an inflection point where technological progress translates into commercialization.
This financing will accelerate that progress, strengthening the system’s ability to formulate research questions, test hypotheses and improve autonomously as the company advances toward superintelligence in life sciences.
Investor Perspectives
China Life Capital said:
“We are pleased to lead BioGeometry’s Series A+ financing. AI for Science (AI4S) is transforming scientific research and creating new sources of innovation-driven productivity, making it a major frontier of global technological competition. Leading this round is a strategic investment in AI4S’s potential to reshape life sciences R&D worldwide. It also advances our commitment to applying AI in high-value settings and developing new drivers of productivity. As a leading AI-native biotechnology company in China, BioGeometry has established distinctive technological advantages through its molecular and cellular world models and recursive self-improvement (RSI) framework linking computational design with wet-lab experimentation. Its industry partnerships and proprietary pipeline are increasingly translating these capabilities into value. China Life Capital will combine its industry resources and investment capabilities to support BioGeometry, advance the integration of AI and life sciences, and accelerate the adoption of AI across China’s biopharmaceutical industry.”
Qiushan Guo, Partner at the Shanghai Biomedical Innovation and Translation Fund, said:
“From co-leading BioGeometry’s first financing round to investing again in this round, our investment conviction has remained consistent and has been reinforced over time. The company’s modeling capabilities now span molecules and cells. GeoFlow models molecular interactions at atomic resolution, integrating structure prediction, de novo design and multistep reasoning. GeoCell models cellular transcriptomic responses to perturbations, extending the company’s capabilities from molecular design to disease biology across target discovery, indication selection and patient stratification. The RSI framework links computation and wet-lab experimentation, enabling the models to improve through real experimental feedback and strengthening the company’s technological advantages as data accumulate. Several bispecific antibody programs are also advancing toward the IND stage. Our follow-on investment reflects our confidence in the team’s execution and in the long-term value of this approach. We look forward to BioGeometry building on its rapidly evolving GeoFlow platform, developed and controlled in-house, to accelerate the clinical development of its proprietary pipeline and expand global partnerships.”
About BioGeometry
BioGeometry is an AI-native biotechnology company. It uses generative AI to build biological world models for life sciences: GeoFlow at the molecular scale and GeoCell at the cellular scale. Its established RSI framework connects computational modeling with wet-lab experimentation to advance autonomous scientific discovery across antibodies, vaccines, peptides and enzyme engineering. The company aims to progress from understanding biological systems to designing them. BioGeometry is headquartered in Beijing and has an R&D center in Shanghai.
Business inquiries: bd@biogeom.com
Media inquiries: pr@biogeom.com