Invest2 publishers3 min readPublished
Novo Nordisk took equity in the buyer of the Parkinson's therapy it shelved
Cellular Intelligence has added Bob Langer and Yann LeCun to its scientific advisory board. The $62m to $70m it has raised now has to take STEM-PD, the cell therapy Novo Nordisk dropped, through Phase 2.
The Investor · Invest desk
What happened
- Cellular Intelligence said Monday that Bob Langer and Yann LeCun had joined its scientific advisory board, with Jens Nielsen of the BioInnovation Institute and Fabian Theis of Helmholtz Munich.
- The Boston company was known as Somite AI until January 2026, when it rebranded and published a white paper setting out what it calls cell fate engineering.
- Its largest round is a $47m Series A from May 2025 led by Khosla Ventures, with the Chan Zuckerberg Initiative and AMD Ventures among the participants.
Compiled by The InvestorSomething wrong?How this is made
Why it matters
- constraint Between $62m and $70m has to pay for the capsule data engine, the foundation model and a Phase 2 trial at the same time, so money spent generating more signal combinations is money not spent dosing patients.
- exposure Cellular Intelligence's shareholders now carry the cost of a Parkinson's trial that Novo Nordisk stopped funding, while Novo keeps a claim on the result through its equity stake.
- precedent A large pharma company handing a discontinued cell therapy to a startup and taking equity back gives other shelved programs a route out that costs the seller little cash.
- contradiction Fortune has LeCun leaving Meta last December to run AMI Labs; Crypto Briefing still calls him Meta's chief AI scientist. Which is current changes how much time a new adviser has to give.
Novo Nordisk did two things in the same May 2026 transaction. It gave up global rights to STEM-PD, an allogeneic Parkinson's cell therapy that it had shelved, and it took a strategic equity position in Cellular Intelligence, the company that picked the program up [10][11]. Cellular Intelligence now pays for the clinical work while Novo holds equity in the outcome. Neither Fortune nor Crypto Briefing reported the price of the asset or the size of the stake.
The balance sheet carrying that program is between $62m and $70m. The $47m Series A that Khosla Ventures led in May 2025, with the Chan Zuckerberg Initiative and AMD Ventures participating, is 67% to 76% of that [12][13][14]. That round closed 16 months before the board announcement [15]. The same money funds three things at once, which Langer described to Fortune as "a proprietary engine for generating biological data over time, an AI foundation model built to learn from that data, and an owned clinical program where it can test the practical value of its predictions" [16].
On the data side, Crypto Briefing reports the platform has gone from 27,000 experimental conditions to more than 1 million unique sequential signal combinations feeding transformer-style models, roughly a 37-fold increase. The capsule technology reportedly runs at more than 1,000 times the throughput of conventional methods [6][8][7]. The prediction results so far are internal. Langer told Fortune that early benchmarks beat existing models in both healthy development and cancer, "despite never training on a single cancer cell" [17].
Micha Breakstone, the chief executive, described the product to Fortune in world-model terms: "A world model learns how a system behaves and predicts what happens when you act on it" [18]. He added, "For us, the system is a living cell: the basic computational unit of life, receiving signals, processing information and changing its behavior in response" [19].
Reading four advisory seats as a capital signal needs care. Langer has co-founded roughly 40 companies and holds more than 1,500 issued and pending patents [20]. His name appearing on a cell-therapy startup is a frequent event; the observer seat on the board of directors is the less frequent part [4]. Breakstone also credited Fidji Simo, OpenAI's chief executive of applications, with shaping strategy as an advisor and consultant [26]. Langer told Fortune: "What matters to me is whether the science can lead to something that makes a difference for patients, and whether the people involved can carry it through" [21].
The announcements have come fast: renamed from Somite AI in January 2026, STEM-PD acquired in May, Arjun Raj named chief scientific officer in August, four advisers in September [2][10][24][1]. Crypto Briefing puts the company's focus on cell signaling dynamics, against Recursion Pharmaceuticals, Isomorphic Labs and Insilico Medicine working on molecular structure and drug-target interactions [25].
The two publishers do not agree on LeCun's employer. Fortune says he led Meta's Fundamental AI Research lab as chief AI scientist from 2013 to 2025 and left last December to launch AMI Labs, where he is executive chairman [22]. Crypto Briefing calls him Meta's chief AI scientist [23].
In my view the equity-for-asset structure tells you more than the four names do. Novo stopped paying for STEM-PD and kept a claim on it. That is a cheap option for the seller and a funded obligation for the buyer. If Novo's stake turns out to be large against the $62m to $70m raised, that read flips and the seller is co-funding the program it discontinued.
What to watch
- Whether Cellular Intelligence raises again before Phase 2 dosing begins, and at what valuation against the $47m Series A.
- Whether the prediction benchmarks Langer described turn up outside the company's own white paper and internal results.
- Whether Novo Nordisk discloses the size of its equity stake, or moves other discontinued programs on the same asset-for-equity terms.