Home Exclusive: Targeting General-Purpose Cell Models, Yuanshi Technology Closes Two Funding Rounds in Succession

Exclusive: Targeting General-Purpose Cell Models, Yuanshi Technology Closes Two Funding Rounds in Succession

Oct 22, 2026 07:59 CST Updated 14:39
Sinovation Ventures

Technology-Focused Venture Capital Firm

Shanghai Engine Fund

Early-Stage Investment Firms

Yanyuan Fund

Early-Stage Venture Capital Firms

Monolith Management

Investment Management Institution

Dalton Venture

Venture Capital Institutions in the Medical and Health Field

YUANSHI TECHNOLOGY

AI Drug Discovery Digital Simulation Software Developer

VCBeat has learned that Yuanshi Technology recently announced that it has closed an angel round and an angel+ round in succession, raising close to RMB 100 million. The angel round was led jointly by Sinovation Ventures, the Shanghai Engine Fund, and YANYUAN FUND; the angel+ round was led by Monolith Management, with Dalton Venture following. The proceeds will mainly fund model R&D, the buildout of data and experimental platforms, and application validation in drug discovery scenarios.


Yuanshi Technology is an early-stage company focused on AI virtual cells (AIVC); it builds a vertical life-science foundation layer around a "biological raw-data engine plus general-purpose cell model" approach. The core question it addresses is how to make AI predict cells' responses more accurately under different disease states, experimental conditions, and interventions, so as to build a general-purpose cell foundation model that continuously understands cells, reasons about cells, predicts cells, and evolves itself through experimental feedback.

 

From Molecular Design to Response Pattern Mining: The Bottleneck in AI Drug Discovery Is Shifting

 

Over the past five years, the main thrust of AI drug discovery has centered on single-point tasks such as molecule generation and structure and property prediction. But as models' capabilities on individual tasks have advanced, the problems have become apparent: drug failures often stem not simply from a molecule binding poorly, but from the difficulty of accurately predicting how cells, tissues, and the body will respond to a drug. The bottleneck in AI drug discovery has shifted from the single task of molecular design to zero-shot generalization prediction of disease responses, that is, the ability to predict across contexts such as new patient populations, new tissue environments, and new perturbation combinations.


This is precisely the problem that general-purpose cell models aim to solve: building a data engine from high-quality, large-scale, continuously accumulating perturbation data; learning the patterns of cellular responses with a cell world model; and then, through experimental validation and feedback, bringing the model ever closer to real biological systems. Yuanshi Technology hopes that a general-purpose cell model with cross-context zero-shot generalization will move AI drug discovery from single-point analysis toward broader, higher-dimensional prediction.

 

From Data and Models: Building a Continuous Loop of Cell Understanding, Reasoning, and Prediction

 

Yuanshi Technology is building an R&D system composed of model analysis, experimental validation, and feedback to create a self-evolving virtual cell system. In this system, the model learns patterns of cellular response from existing data and generates predictions for new research questions; experiments validate the model's predictions and fill in information that current data do not yet cover; and the validation results are then used to update the model and adjust the direction of subsequent research. Model, experiment, and feedback form a continuous loop.

 

Why Investors Chose Yuanshi Technology

 

Ren Bobing, General Manager of the Frontier Technology Fund at Sinovation Ventures, said: "AI4S has always been a focus for Sinovation Ventures. We have made systematic investments across AI drug discovery and AI4S infrastructure in projects including Insilico Medicine, Syneron Bio, RiboPeutic, MegaRobo Technologies, and ChemLex. Yuanshi Technology is another important AI4S investment in our progression from AI drug discovery to AI systems biology.


With Insilico Medicine's successful listing on the Hong Kong Stock Exchange, we have already validated the commercial path of AI-driven drug discovery. In terms of technological evolution, 'AI + biological sequences' was the core direction we first anchored. Since AlphaFold2 broke through protein structure prediction in 2021, we have systematically built AI capabilities for understanding and designing biological sequences along the central dogma (DNA to RNA to protein).


But AI drug discovery is only the starting point. From AI drug discovery to AI sequence models and then to AI systems biology, there is a clear path of capability advancement. By 2023 we had upgraded our perspective from a single sequence dimension to 'AI + systems biology' and begun to focus closely on virtual cells. The causal data of virtual cells is one of the keys for biological foundation models to move from 'correlation fitting' to 'mechanism generalization.'


Existing biological foundation models mostly remain at a weakly causal level; observational data alone make it difficult to learn the mechanisms of intervention responses. For this reason, virtual cells are an unavoidable link on the path to a digital human, and the next holy grail of AI biology. They are closer to real physiological states than sequence models, but the demands on data have also undergone a qualitative leap.


Since 2025, a number of virtual cell startups have emerged in Europe and the United States, and industry consensus has formed quickly. We judge that in-house data collection capability is critical to the iteration accuracy and clinical translation value of virtual cell models, and that only high-throughput, low-cost, high-quality cell perturbation data can drive a genuine dry-wet loop. From the perspective of next-generation AI drug discovery, the causal data of virtual cells is changing the core logic of drug discovery, from 'screening compounds' to 'predicting in silico the impact of interventions on cell states.'


The next generation of virtual cells will integrate spatiotemporal transcriptomics with multi-agent systems to close the loop from target discovery to mechanism decoding, and causal perturbation data is the starting point of that loop. Based on this understanding, we invested in Yuanshi Technology, which has a globally leading ultra-high-throughput cell perturbation technology platform that can continuously generate the core experimental data the model needs, providing the key data fuel for virtual cells to move from concept to verifiable, iterable real models."