Shanghai State Capital Invests Hundreds of Millions in AI-Powered Protein Design Firm BioGeometry

Recently, the leading domesticAI for Science (AI4S) EnterprisesBioGeometry（BioGeometry) Announces CompletionNew RoundHundreds of millions of yuan in strategic financing.

This roundFinancingbyShanghai BiopharmaceuticalsInnovation TranslationFunds, Guoke Investment,Fortune Capital,StarLink CapitalCo-lead investor,Co-invested by Gaorong Capital and the Index AI Industry Innovation Fund; Index CapitalServe as the exclusive financial advisor.

This financing round is not only an endorsement from the capital market, but also sends a clear signal: thisDomestically rare capability in foundational large models for the life sciences sector from0-to-1 Capability BuildingThe team has completed technical and industrialization validation, officially ushering in the golden inflection point for rapid commercial growth.

“AI-Native” Leading Enterprise

Building a General-Purpose Operating System for the Microscopic World from Scratch

Unlike the mainstream in the industry,The “AI + Bio” Approach: Viewing AI as a Tool for Localized Efficacy Enhancement，BioGeometry Chooses to Reconstruct the R&D Paradigm of Life Sciences from the Ground Up，Build the Next GenerationAI-Native Biotechnology Company。Amid the current wave of technological advancement, artificial intelligence is accelerating its evolution along three clear main trajectories.：

NumberAI (Digital AI)，Represented by large language models and multimodal models, reconstructing the human world of information and knowledge;

PhysicsAI (Physical AI)，Represented by autonomous driving and humanoid robots, reshaping the physical and mechanical world of humanity;

LifeAI (Bio AI) —— BioGeometry: Solving the Ultimate Question—It is not limited to digits and mechanics; rather, it aims to deeply understand the underlying language of life and achieve programmable, engineered modifications of life at atomic precision.

This is also BioGeometry's and other domestic "The Essential Distinction of “AI + Bio” Companies—While the Industry Relies on Fine-Tuning Foreign Open-Source Models and Remains Stuck in Superficial Application-Layer Patches, BioGeometry Demonstrates Its Robust Hard Power Through Full-Stack Independent R&D Integrating Wet and Dry Labs.

The confidence to build a foundational engine “from 0 to 1” stems from its powerful research team: founded by renowned AI4S scientist Professor Tang Jian, with Turing Award laureate and “father of deep learning” Yoshua Bengio serving as Chief Scientific Advisor, while other team members have achieved multiple international milestone accomplishments in fields such as graph machine learning and diffusion generative models.

Recent, asCore Contributors Participate in NVIDIA’s Open-Source Large Protein ModelR&D of La Proteina, its independently developedAI Virtual Cell Model PerturbDiffIt has garnered widespread attention in the international scientific research community and possesses the robust capabilities to define the next generation of life science infrastructure.

Self-developed“Microscopic World Model”

GeoFlow Leads Global Technological Development Trends

Whether we can truly understand and precisely design intermolecular interactions at the microscopic level is the most fundamental proposition in life sciences.

BioGeometry's solution is to create a system that understands the laws of life and matter.“Microscopic World Model” — The GeoFlow Series. From the generational evolution of GeoFlow, it is clearly evident that BioGeometry continues to lead in technological prowess and firmly maintains its position in the top international tier:

GeoFlow Evolution Milestones：

GeoFlow V1 (2024.06) ── Proprietary Model’s Structure Prediction Capability on Par with AlphaFold 3

GeoFlow V2 (2025.04) ── The world’s first unified “structure prediction + de novo design” platform, achieving a zero-to-one breakthrough in domestic antibody de novo design

GeoFlow V3 (2025.10) ── First introduction of the “multi-step reasoning” closed loop, with a nearly 100-fold surge in hit rates for lead molecules

1.The Generational Advantage of All-Atom Modeling (GeoFlow V1）

Mainstream ProteinsMost AI models treat proteins as linear sequences, performing approximate modeling at the amino acid residue level. In contrast, GeoFlow directly provides a precise characterization of three-dimensional space at the atomic level, incorporating the spatial coordinates of every atom and chemical bond angles into a comprehensive model. In the task of predicting protein-protein complex structures, GeoFlow V1 achieved performance on par with AlphaFold 3 upon its release, establishing its underlying technology among the top tier internationally.

2.Breaking the Global Deadlock: The Perfect Unity of Prediction and Design (GeoFlow V2）

Previously, even as strong asAlphaFold also leans toward “structure prediction,” whereas designing new molecules often requires switching to other tools. GeoFlow V2 is the first model globally to unify protein structure prediction and de novo design within a single framework, becoming China’s first AI large model to achieve breakthroughs in antibody de novo design, thereby directly breaking the monopoly held by overseas closed-source models.

3.Leading the WorldAI4S Large Model Trends: Introducing “Multi-Step Reasoning” for Extended Thinking (GeoFlow V3)

Like large language models (such asOpenAI o1) Evolving towards long-horizon thinking and reasoning, BioGeometry is the first to bring this cutting-edge trend into the microscopic world. GeoFlow V3 breaks through the traditional linear "generate-and-filter" workflow by establishing a multi-step reasoning loop of "generation-evaluation-optimization," enabling autonomous evolution that mimics the natural process of antibody affinity maturation. In de novo design tasks across more than 20 targets, GeoFlow V3 achieved an average hit rate for hit molecules close to 20%, representing an nearly hundredfold improvement over previous-generation AI methods and compressing the lead molecule discovery cycle to under three weeks. This pace of evolution is setting the direction for the global development of AI in life sciences.

Currently, BioGeometry is developingNext-Generation Microscopic World ModelGeoFlow V4, expanding the modeling scale from molecular interactions to the more complex cellular level—shifting from “designing individual molecules” to “designing molecular systems”—and laying the foundational AI infrastructure for next-generation drug discovery and bioengineering.

Breaking Through with Hardcore Cases“The Toolification Curse of ‘PowerPoint Drug Development’”

BioGeometry through“Cooperative Development+Self-Developed Pipeline“Dual-wheel drive has delivered multiple industry-shattering cases in high-difficulty fields such as biopharmaceuticals and synthetic biology, breaking through industry ceilings.”

Biopharmaceuticals: Reshaping Antibody Drug and Vaccine Development

• Tumor Immunology—De Novo Design of High-Specificity Antibodies: Confronting proteins within homologous families that are highly similar and indistinguishable by traditional methods, GeoFlow directly incorporates "specificity" as a prior constraint into the generation phase. By designing no more than 100 sequences, it successfully identified two antibodies exhibiting both high selectivity and high affinity, thereby pushing the boundaries of existing AI tools.

  
  

• International Pharmaceutical Collaboration——Lead Antibody Multi-Parameter Simultaneous Optimization: For a multi-objective optimization project that an internal team at a renowned foreign pharmaceutical company failed to resolve after more than a year of effort, GeoFlow delivered perfect antibody molecules with affinity improved by dozens of fold, expression levels increased 8-fold, and humanization optimized to over 90%, all through a single design cycle in a zero-shot setting (without fine-tuning on target-specific data). The project delivery timeline was shortened by more than 80% compared to the client’s expectations.

  
  

• Vaccine Development——From “Naturally Impossible” to “Engineering Feasible”: Addressing the long-standing industry challenge where a certain virus’s native antigenic protein fails to stably form dimers, GeoFlow increased the dimer proportion of this protein from less than 10% to over 90%, while simultaneously boosting expression levels by dozens of times, thereby fundamentally overcoming the limitations of the native structure.

Synthetic Biology: Redesigning Industrial Enzymes, with Multiple Pipelines Completing Pilot-Scale Amplification

• Natural Borneol: World’s First Biological Synthesis Achieved, with Chiral Purity Reaching Up to99.9%, with an 80% significant reduction in cost compared to traditional plant extraction

  
  

• α-Ketoglutarate: Targeted optimization of key enzyme activity reduces costs by over 60% compared to existing market biosynthesis technologies

From Molecules to Cells: From Understanding Life to Designing Life— BioGeometry is building the foundational infrastructure that enables life to be engineered and written.

Revaluation:TechnologyR&DWith Businessof the intersection of chemical"Outbreak Inflection Point"

According to insiders, dozens of top-tier institutions competed to invest hundreds of millions of yuan in this funding round, with the core reason being that BioGeometry has already completed its transition fromThe leap from a “technological high ground” to an achievable “commercialization high ground.”

Among them, multiple investment institutions stated that “BioGeometry is one of the few native (AI Native) definers of underlying infrastructure in the global life sciences field,” “a key breakthrough in breaking the monopoly of overseas closed-source models,” “highly recognizing the team’s technical capabilities and global influence in the AI4S domain,” and “with an increasingly clear business model, the company may be approaching a valuation inflection point.”

The injection of hundreds of millions in financing this round has officially pressed the fast-forward button on BioGeometry’s exponential valuation growth. The development of the next-generation GeoFlow V4 model will extend the modeling scale from “single molecules” to “cellular-level molecular systems,” further amplifying the company’s technological barriers and commercial ceiling.

Guo Qiushan, President of the Shanghai Biomedical Innovation and Translation Fund, stated，The development of macromolecular drugs has long been constrained by the protracted nature of traditional screening and the previous stepwise approachError amplification cascades through AI toolchains. BioGeometry has achieved atomic-level precision in modeling biomolecular interactions, integrating structure prediction, sequence generation, druggability assessment, and wet-lab feedback into a closed loop, representing an approach more aligned with fundamental scientific logic and real-world industrial needs. AI Native Pathway.

This all-atom De Novo Design philosophy, enabling the company to demonstrate unparalleled generational advantages in high-difficulty pipelines such as traditionally challenging drug targets, complex antibodies, and multispecific macromolecules, and achieving PCC delivery of grade-level molecules. We look forward to BioGeometry leveraging its rapid iteration capabilities and self-controllable, GeoFlow Algorithm foundation, accelerating the clinical development of self-developed pipelines and global collaboration at full speed.

Zhang Kun, Head of the Smart Healthcare Group at Guoke Investment, stated:AI-driven drug discovery is poised to break the R&D bottleneck posed by “Eroom’s Law” in the biopharmaceutical industry, fostering a new paradigm for macromolecular therapeutics characterized by structural understanding, targeted design, and closed-loop validation through integrated dry- and wet-lab experiments. We highly recognize the technical expertise and global influence of Professor Tang Jian’s team in the field of AI for Science (AI4S). Their independently developed GeoFlow model has already demonstrated distinct technological advantages in the de novo design of antibody drugs and industrial enzymes. We believe that, against the backdrop of AI reshaping drug development, BioGeometry will deeply empower innovative pharmaceutical companies and the biomanufacturing industry, accelerating pipeline translation and commercialization, thereby continuously delivering industrial value and realizing its commercial potential.

Dr. Wang Dakui, Managing Director of Fortune Capital, stated:The “emergent intelligence” moment for AI in the biopharmaceutical sector has arrived faster than industry expectations. AI can design and screen candidate molecules at a nearly infinite scale, shifting drug discovery innovation from traditional trial-and-error experimentation to computation-driven approaches. BioGeometry, led by Professor Tang Jian, is a top-tier AI4S team deeply rooted in biological computing. The team boasts solid academic foundations and the capability to translate cutting-edge algorithms into practical engineering solutions. Its self-developed GeoFlow model of the microscopic world can predict the structures and interactions of biomacromolecules, such as proteins, with high precision. This technology places BioGeometry among the global leaders and serves as a key breakthrough in breaking the monopoly of overseas closed-source models. Coupled with China’s advantages in wet-lab experiments—namely lower costs and faster iteration cycles—Chinese AI-driven pharmaceutical companies represented by BioGeometry are fully capable of achieving late-mover catch-up.

Li Wenjue, Partner at Xinglian Capital, stated that life sciences are entering a new era: moving from reliance on experience and serendipitous discoveries toward precision innovation driven by computation and design. BioGeometry leverages generativePowered by AI, we explore the programmable design of proteins—the fundamental language of life—and continuously accelerate model iteration and experimental validation through a closed-loop integration of computational and wet-lab experiments, thereby enhancing the efficiency and success rate of novel molecule discovery and functional design. We are optimistic about BioGeometry’s systematic advancements in AI foundation models, protein design capabilities, and experimental validation systems, as well as the long-term innovative potential demonstrated by its global, interdisciplinary team. We look forward to BioGeometry’s continued promotion of the deep integration of AI and life sciences, ushering in a new paradigm for biopharmaceuticals and synthetic biology that is more efficient, predictable, and engineering-driven.

—The End—

Recommended Reading