Gero.

01 // Who They Are NowThe longevity startup that learned to speak pharma

Gero looks, on paper, like a small biotech. About 140 people. Roughly eight million dollars in disclosed venture capital. A website that prefers diagrams to slogans. Then you look at who they are signing deals with - Pfizer on fibrotic disease, Chugai Pharmaceutical on antibody therapeutics for age-related conditions - and the proportion tips. In July 2025, Chugai committed to a joint research and license agreement worth up to roughly 250 million dollars in milestones if everything goes well. For a Series A company, that is not a partnership. That is a verdict.

The verdict, condensed, is this: Gero may have built something that the rest of biotech has been searching for. A foundational model of human health, trained on more than one hundred million longitudinal medical records, that can tell the difference between a disease you can still fix and a process - aging - that requires fixing in a different way.

02 // The Problem They SawMost biotech AI is trained on the wrong thing

Modern drug discovery has a famous embarrassment. Hundreds of compounds extend the lifespans of mice. Almost none extend the lifespans of humans. The data that pharmaceutical companies use is, with the politeness of an industry that does not like to admit anything, narrow - cell cultures, model organisms, clinical trials of a few thousand people lasting a few years. Aging, meanwhile, happens to billions over decades.

Gero's founders saw the gap and chose, with a degree of stubbornness that turns out to be the through-line of their story, to ignore it. Their argument: if you want to model what happens to a human over fifty years, you need fifty years of real human data, and you need a framework borrowed from somewhere that already takes complex systems seriously. Physics, for example.

It is the kind of bet that sounds reasonable in a TED talk and impossible in a lab. Most labs do not have the data. Most physicists do not have the biology. Most AI startups do not have either - they have a transformer and a press release.

03 // The Founders' BetA physicist walks into a biology lab

Peter Fedichev, Gero's CEO, has a PhD in theoretical physics. Before Gero, he co-founded Quantum Pharmaceuticals, where he tried, with mixed but instructive results, to apply quantum mechanics to drug design. Maxim Kholin, his co-founder, came from the entrepreneurial side - the side that has to make payroll while the physicist is rewriting the laws of biology in a notebook.

Together, in 2018, they made a bet that the right way to understand aging was to look at it as thermodynamics - a creeping rise in disorder that follows describable, sometimes elegant, mathematical rules. They convinced Bulba Ventures to write a check, then Melnichek Investments, then VitaDAO, and built a small but dense team in Singapore. The headquarters choice mattered. Singapore has invested heavily in longevity research - Brian Kennedy, a distinguished professor at the National University of Singapore and one of the field's most cited voices, eventually joined Gero's board.

The bet was not subtle. If aging obeys physics, then aging can be measured. If aging can be measured, then drugs that slow it can be tested. If drugs that slow it can be tested, then the entire pharmaceutical pipeline - currently optimized for diseases that arrive one at a time - can be redirected at the thing that produces most of them.

04 // The ProductWhat a foundational model of human health actually does

Strip away the slogans and Gero's core product is a piece of software with an unfashionably long name and an unfashionably specific job. It is a foundational health model - meaning it has been trained, in the manner of a large language model, but on medical records instead of internet text. The training data is more than one hundred million longitudinal records: blood tests over years, diagnoses over decades, the small revealing drift of biomarkers as a person ages.

The output, given a person's history, is not a diagnosis. It is a trajectory. The model can identify which parts of someone's health are reversible disease - the kind a drug can fix - and which parts are the slower thermodynamic creep of aging itself. That distinction is, for a pharmaceutical company hunting for a target, the entire game.

Gero then translates the targets it identifies into something a pharma partner can build a drug around. In 2025 the company released ProtoBind-Diff, a generative AI model that designs drug candidates directly from protein sequences without needing the protein's structure first. That is, again, a small sentence with a long implication. Structure-based drug design is the standard. Structure-free design is faster, cheaper, and possible only if your model knows the physics well enough to fill in what is missing.

Fig. 1 - The unglamorous spreadsheet behind an unreasonably ambitious mission.

05 // The ProofCustomers, data, and the company you keep

You can tell a great deal about a biotech by reading its partner list out loud. Gero's list reads like an attempt to embarrass anyone who thinks they are just a Singapore startup. Pfizer for fibrotic disease. Chugai - the Japanese pharma giant majority-owned by Roche - for age-related disease antibodies. Foxo Technologies for epigenetic biomarkers. The National University of Singapore, Harvard Medical School, and Roswell Park as academic anchors. Brian Kennedy, Vadim Gladyshev and Andrei Gudkov as advisors, names that anyone reading this for a second time in a longevity context will recognise.

The cards below are the kind of thing that, three years ago, would have read like a manifesto. Today, with a 250 million dollar check at the end of one of them, they read more like a balance sheet.

06 // The MissionDouble the human healthspan. Within one generation.

Gero's stated mission is, depending on your temperament, either modest or insane. To eliminate the root causes of age-related diseases. To double human health- and lifespan within the current generation. The company is too earnest to call it a moonshot and too well-funded to call it a fantasy. It is, however, the kind of sentence that requires a footnote, and the footnote is: nobody else is closer.

The mission is also a redirection of an entire industry. Pharmaceutical companies traditionally pick a disease and chase its target. Gero picks a person and chases the underlying clock. If they are right - and the partner list suggests at least some of the right people think they might be - the consequence is that a single platform could feed multiple disease pipelines instead of running on the one-drug-one-disease economics that has defined biotech since penicillin.

07 // Why It Matters TomorrowThe quiet decade ahead

Drug development is slow on purpose. Gero will not have a marketed therapy in 2026, or 2027, or possibly for several years after. The Chugai partnership, like all such partnerships, runs on milestones rather than calendars. What happens in the meantime is, in some ways, more interesting. The foundational model keeps training. New collaborations get announced. New targets get found and licensed. The platform becomes, with each cycle, harder to dismiss.

That is the real value of a foundational model in a pharmaceutical context. You do not need to be right about every drug. You need to be right often enough that the people licensing your targets keep coming back. Gero is, by any reasonable read, beginning to build that habit.

And so it is May 2026 again, and the model is still running. The records are still being added. Somewhere in the noise, a target is being identified that a pharma partner will, two or four or seven years from now, turn into something a doctor prescribes. The Singapore office is still smaller than the ambition. The Palo Alto address is still a useful piece of geography. Peter Fedichev is still a physicist and still, by his own admission, trying to do something the field told him was a category error.

The category error is starting to look like a category.

08 // Watch & ReadInterviews and product demos