Startups & Funding

INFevo Raises Over RMB 100M to Build AI Virtual Cells and Robotic Labs

INFevo raised over RMB 100M in a Pre-A round to expand AI virtual-cell models and robotic wet labs, its second round in five months and a record for a Chinese virtual-cell firm.

INFevo Raises Over RMB 100M Pre-A to Expand AI Virtual-Cell Research
INFevo Raises Over RMB 100M Pre-A to Expand AI Virtual-Cell ResearchThis_is_Engineering / pixabay
By Elena Vasquez5 min read

Why it matters

  • INFevo raised over RMB 100 million in a Pre-A round announced November 16, 2026 — its second round in five months
  • Xinglian Capital called it a record single-round financing for a Chinese virtual-cell company
  • Karl Workbench ranked first in single-cell and spatial omics on SciAgentArena in an October 2026 retest running on DeepSeek
  • The OCOO-T model reports state-of-the-art results on Tahoe100M, Replogle and PBMC benchmarks
  • INFevo is working with a Beijing key laboratory led by Peking University First Hospital on liver, biliary and pancreatic tumor organoids

Chinese startup INFevo has raised more than RMB 100 million in a Pre-A round to expand its AI virtual-cell models and automated experimental systems, investor Xinglian Capital announced on November 16, 2026. The round is INFevo's second in five months, and Xinglian described it as a record single-round financing for a Chinese virtual-cell company.

The money signals where Chinese investors think the next phase of AI-for-science is heading: not just models that predict, but closed loops where predictions get tested by robots and the results feed back into the models. The stakes are real for pharmaceutical research, where drug discovery still burns years and billions on hypotheses that fail in the lab.

Who put in the money?

The round drew a mix of new and returning investors:

  • New investors: Xinglian Capital, Prosperity7 Ventures, Huagai Capital, Huakong Fund and DT Capital
  • Follow-ons: Existing backers Shunwei Capital, HSG and Songhe Capital added to their positions

The Pre-A follows an angel round of tens of millions of yuan in June 2026. According to Xinglian's announcement, the fresh capital will fund expansion of the virtual-cell models and automated experimental systems, plus deeper collaboration with pharmaceutical companies, hospitals and research institutions.

INFevo was incubated by XtalPi. In a July 16, 2026 announcement, XtalPi described how its molecular-design and automated-synthesis capabilities could complement INFevo's predictions, forming a chain from molecular generation through cellular-response prediction to experimental validation.

What is a virtual cell, in practice?

INFevo's virtual-cell approach predicts how cells will respond to drug or genetic interventions before anyone runs a laboratory experiment. Researchers could then compare potential targets, rank candidate molecules and save wet-lab capacity for the questions that look most promising.

The technical core is described in the company's OCOO-T paper. It is a minimalist flow-matching model whose standard Transformer stack operates directly on continuous gene-expression profiles. Prediction is formulated as continuous-time denoising. Perturbation, dosage and cell-line or cell-type information enter through adaptive layer normalization and in-context tokens. The model also scales to long transcriptional profiles by patching and depatching cellular contexts.

The authors report state-of-the-art results across three benchmarks: Tahoe100M, Replogle and PBMC, covering chemical, genetic and cytokine perturbations.

Beyond a single model, INFevo is building what it calls recursive scientific discovery. The system retains scientific questions, hypotheses, analytical methods and experimental results so they can inform later hypotheses, experiment designs and candidate selection. Its automated wet-lab platform covers cell culture, perturbation and multi-omics measurement. Model predictions help choose which experiments to run; the resulting data test those predictions and feed subsequent model improvements.

What does the Karl Workbench actually do?

The hypothesis side of the pipeline belongs to the Popper Project, named after philosopher Karl Popper. It uses research agents and professional databases to generate, challenge and refine testable hypotheses, displaying traceable reasoning chains, evidence gaps and proposed experiments.

Its tool, Karl Workbench, has entered beta testing. It includes research records, accumulated evidence, outputs, supervision and iterative-improvement functions that let researchers review the system's judgments and resume earlier tasks.

Xinglian reported that an October 2026 retest using SciAgentArena placed Karl Workbench first overall in both single-cell and spatial omics when running on DeepSeek. SciAgentArena comprises roughly 200 stepwise-verified tasks spanning drug discovery, single-cell omics, spatial omics, electronic health records and genetics. It separates agent execution from evaluation and assesses data analysis, optimization, discovery and validity checking.

Why the benchmark win needs context

SciAgentArena's own authors found that no single agent dominated across tasks and domains. Current systems, they reported, performed unevenly, struggled with novel discovery and self-directed exploration, and tended to execute requests even when the scientific premise behind them was invalid.

That wider record matters for reading INFevo's result. The retest covers Karl Workbench running on DeepSeek in two omics categories. It does not establish a general lead across scientific research. A first-place finish on a benchmark whose authors document uneven agent performance is a useful signal, not a verdict.

This is the recurring tension in AI-for-science: developer-reported benchmark scores arrive quickly, while validation in real research workflows arrives slowly. SciAgentArena's findings about agents executing invalid premises are exactly the failure mode that closed-loop systems like INFevo's are designed to catch — if the wet-lab side works as advertised.

Where the clinical test begins

INFevo is also participating in a Beijing key laboratory led by Peking University First Hospital. The lab focuses on liver, biliary and pancreatic tumor organoid culture and intelligent diagnosis and treatment. Its stated aim is to connect clinical questions, patient-derived organoid models, drug screening, AI analysis and clinical validation.

That collaboration puts the virtual-cell system in direct contact with patient-derived clinical samples. Model predictions and organoid experiments are intended to inform one another, which would test whether the approach survives contact with the messiness of real tumors rather than clean benchmark datasets.

Co-founder Yalong Zhao said the financing would support the foundation-model and automated-experiment infrastructure and accelerate the recursive-discovery system. Another co-founder, Lipeng Lai, also serves as XtalPi's co-founder and chief innovation officer — a sign of how tightly INFevo's ambitions tie into XtalPi's molecular-design and automated-synthesis stack.

What comes next

The new capital gives INFevo resources to expand both halves of its system: computational prediction and automated validation. The consequential test is whether developer-reported benchmark results and beta research tools can become repeatable performance on patient-derived samples and inside pharmaceutical research workflows. Two funding rounds in five months bought the startup time and hardware; the organoid lab in Beijing will tell it whether the loop actually closes.

Source: unite.ai

Share this article:

More from Elena Vasquez

Elena Vasquez

Show full bio

Market editor covering technology and product news in artificial intelligence and machine learning at AI In Context.

708 articles

Related articles

  1. Isomorphic Labs raises $5 billion, its second mega-round in six months
  2. Korea's AI Funding Boom Concentrates on Chips and Models, Starving Services
  3. Manus Raises Over $500M in First Round Since Meta Deal Collapsed
  4. Isomorphic Labs raises $5B to push AI-designed drugs into trials
  5. Manus Raises Over $500 Million in First Round Since Blocked Meta Deal

« Previous articleNext article »