The company emerged from stealth Tuesday with seed funding to develop what it calls an "AI Virtual Cell" that predicts cellular behavior without running every lab experiment.
Rivercell, a Paris-based startup led by veterans of AI-driven drug development, announced on October 7, 2026, it had secured $25 million in seed financing led by HV, with backing from HCVC, Alven and Bpifrance Digital Venture. The round was structured entirely as equity, according to French business publication Les Echos. The company declined to share its valuation.
The capital will fund an ambitious effort to build what Rivercell describes as a foundation model for cellular biology, one designed to forecast how human cells react to pharmaceutical compounds or genetic modifications. At the core of the startup's pitch is a marriage of automated laboratory infrastructure with machine learning systems that decide which experiments to conduct next, generating the kind of training data that has eluded much of the AI drug discovery field.
"A world model of the cell is the opportunity of the century in medicine," said Yann Fleureau, co-founder and CEO, in a statement accompanying the announcement. "The missing piece is data."
Hardware Meets Software
Rivercell plans to use the proceeds to expand its proprietary wet lab in Paris and launch its AI Virtual Cell initiative. The company frames its integrated hardware-software platform as a "GPU for the bio data center," a reference to the specialized chips that power machine learning workloads.
The technical approach involves running what researchers call interventional experiments: subjecting cells to drug treatments, genetic edits or environmental shifts, then capturing live-cell imaging throughout the process and molecular snapshots at the endpoint. All on the same cells. The system is built to scale across different cell types and genetic backgrounds, with an AI model that tracks a probabilistic "belief state" about each partially observed cell and selects the next round of informative tests, according to a manifesto the company published on its website.
Existing public datasets, Rivercell argues, fall short because they are largely observational, limited to a single type of measurement, and static. That makes them poor fuel for models trying to predict how cells shift under specific conditions across varied contexts.
Experienced Operators
Fleureau previously co-founded Cardiologs, an AI-based cardiac diagnostics firm that Philips acquired in 2021. He studied at Ecole Polytechnique and UC Berkeley. His co-founder, chief scientific officer Eric Durand, brings a résumé that includes leading oncology data science at Novartis, serving as chief data science officer at Owkin, and co-founding Bioptimus, which had raised $76 million as of January 14, 2025. Durand earned a PhD in statistics and population genetics and completed postdoctoral work.
The startup has assembled a scientific advisory board with notable names, including Charlotte Bunne, a tenure-track assistant professor in artificial intelligence in molecular medicine at EPFL, and Fabian Theis, who directs the Computational Health Center at Helmholtz Munich.
A Crowded, Well-Funded Field
The $25 million seed puts Rivercell well above the median for early-stage rounds, which Crunchbase data has pegged in recent years around $3 million to $4 million in the U.S., though AI-focused deals often command heftier sums.
Rivercell enters a space that has seen fresh energy around what some are calling "world models" of cellular behavior. London-based Relation Therapeutics unveiled its cellular foundation model, MORGAN, on July 30, 2026, and later announced an expanded collaboration with GSK valued at up to $110 million, centered on large-scale human cellular perturbation data. Budapest's Turbine, which operates simulated cell platforms for pharmaceutical clients, has also raised significant capital in recent funding rounds.
Rivercell says potential applications stretch across oncology, immunology, rare diseases and cardiometabolic conditions, though the company did not offer timelines for when initial model results might surface.
Building in Paris
Maxi Pethö-Schramm, a principal at HV, said in the release that "the next frontier is predicting how cells behave, and that requires a full-stack approach that combines a novel platform, automated wet labs, and world models."
Durand wrote on LinkedIn on Tuesday that the team is "assembling a world-class team in Paris across biology, hardware, software and ML." The company's careers portal lists openings in thin films, fluidics, biotech and machine learning, signaling plans to scale quickly.
Whether Rivercell can deliver on its vision remains to be seen. But the startup has positioned itself at the intersection of two trends: the industrial-scale automation of lab work and the hunt for AI models that don't just describe biology but predict it. That convergence, investors seem to believe, might be worth betting on.
