Cellular Intelligence, which acquired Novo Nordisk’s abandoned Parkinson’s cell therapy, brings Moderna cofounder Robert Langer and AI scientist Yann LeCun onto the scientific advisory board for its AI model of a living cell
Cellular Intelligence, which acquired Novo Nordisk’s abandoned Parkinson’s cell therapy, brings Moderna cofounder Robert Langer and AI scientist Yann LeCun onto the scientific advisory board for its AI model of a living cell
On September 21, Boston-based Cellular Intelligence (CI) announced that its scientific advisory board includes Langer, LeCun, Jens Nielsen, who heads the Novo Nordisk Foundation’s BioInnovation Institute, and Human Cell Atlas cofounder Fabian Theis. Langer also joined the board of directors as an observer. CI is building an AI model that predicts how a cell will respond to a signal. It already owns a Parkinson’s therapy in clinical development that Novo Nordisk abandoned when it closed its entire cell therapy division in October 2025.
Developing cell therapies involves trial and error: a laboratory tests signaling molecules and their timing until a stem cell becomes the desired cell type. The process takes years and costs tens of millions of dollars. CI’s cofounders, particularly Harvard Medical School geneticist Olivier Pourquié, showed CEO Micha Breakstone, who sold conversational AI company Chorus.ai for $575 million, that about 20 core signaling pathways govern development from a single cell into an entire human body. Whether a cell becomes a neuron, a heart cell, or a pancreatic beta cell depends on the combination, dose, and sequence of these signals. The same logic governs disease.
CI built a capsule-based platform in which stem cell colonies undergo successive rounds of signaling, with each colony’s barcode recording its treatment history. This allowed the company to scale from a pilot testing 27 thousand combinations to more than a million in 2025. It also became possible to distinguish separate skeletal and cardiac lineages within muscle progenitor cells that had previously been grouped together. Given a signal, the model predicts what a cell will become.
In May 2026, CI acquired global rights from Novo Nordisk to STEM-PD, a Parkinson’s therapy with FDA Fast Track designation that is ready for Phase II. STEM-PD data feed into CI’s models: predictions that were previously tested only in cell culture are now also tested in living patients. According to its one-year data, the cells engrafted in seven of eight patients; one patient died from an infection. In October 2025, Novo Nordisk’s new CEO closed the entire cell therapy division to save $1,3 billion a year, cutting about 250 jobs. In exchange for the therapy rights, Novo Nordisk received an equity stake in CI and the right to future royalties.
LeCun, a Turing Award winner who led AI research at Meta for 12 years, left the company at the end of 2025 and almost immediately cofounded AMI Labs to pursue “world models.” He describes his involvement on CI’s advisory board in the same terms:
I am interested in models that learn from data how the world works and predict what will happen when we intervene in it. A living cell is an exceptionally rich instance of that very problem.
In an interview with Fortune, Breakstone describes CI’s platform in the same terms. LeCun is testing the same premise in parallel: in the physical world at AMI Labs and in living cells through his advisory role at CI. Two months earlier, his own fund for investments in AI and healthcare closed on the day it was announced, without ever confirming the capital behind his prominent name. His role on CI’s advisory board puts only his reputation and expertise at stake.
Breakstone told Fortune why the company needs this advisory board: CI plans to license its model to pharmaceutical partners for drug discovery, and such deals require independent assurance of quality. Langer, the most cited engineer in history, holds more than 1500 patents and has cofounded about 40 companies, including Moderna. His reputation provides that assurance, although his role on CI’s board of directors is limited to that of an observer.