Virtual Cell AI Is Booming. Longevity Founders in San Francisco Want Proof First

Why the virtual cell is AI biology's biggest bet

The "virtual cell" has become one of the biggest bets in AI biology. The day before the event, on October 7, Biohub, the nonprofit founded by Mark Zuckerberg and Priscilla Chan, expanded its Virtual Biology Initiative into a $1.8 billion effort, counting new and existing funding, with the US Department of Energy, the National Institutes of Health (NIH), Google DeepMind, Meta and Isomorphic Labs.

The idea is simple to state and hard to build: software that predicts how a human cell will respond to a drug, a mutation or a disease, so scientists can run experiments on a computer before they run them in a lab. The Arc Institute released its first model, State, in June 2025, trained on data from more than 270 million cells. In August 2026, GenBio AI announced AIDO Cell, which it pitches as a general-purpose cell simulator.

But the science is still contested. A 2025 study in Nature Methods found that deep-learning models did not yet beat simple linear baselines at predicting the effects of gene perturbations. A 2026 benchmark from the Shanghai AI Laboratory found that which model looks best depends heavily on the task and the metric.

That tension set the agenda on October 8, when AI Insiders hosted AI × Longevity at Citizens Private Bank at 101 California Street, part of SF Tech Week. The night drew 497 approved registrations from founders, investors and researchers.

Prediction is not proof

The evening opened with a keynote from Ronjon Nag of R42 Group and Stanford Medicine. Then three people who build very different parts of the health stack took the stage: Christin Glorioso, CEO of NeuroAge Therapeutics, which works on brain aging; Jacob Peters, co-founder and chairman of the health-testing company Superpower; and Girish Redekar, CEO of the compliance platform Sprinto. Hannah Tono of the Computational Design Institute moderated.

Their starting point: a model that predicts biology well has not proven anything yet. Before a result reaches a patient, the panel argued, it needs a clear context of use. That means naming the decision it changes, and for which person.

Longevity biomarkers: more data, fewer answers

The panel was blunt about consumer health testing. People can now get back dozens of biomarkers from a single blood draw, and only a few of them should change what anyone does.

A prediction with nothing to act on, they said, can do more harm than good. A worrying number with no next step is not a service to the person who receives it.

The bottleneck moved to experimental data

The biggest shift the panel described is where progress now gets stuck. It's no longer the models. It's experimental data: the lab work needed to check whether a prediction holds, and the clean, consented datasets needed to train the next model.

For founders, that changes what investors will ask. Trust, validation and data quality are now engineering problems, not marketing ones.

The longevity startups the room backed

After the panel, four startups pitched, and guests voted on the most deployable, most investable and most game-changing company.

One company won all three.

AWEAR took first place across the board. It makes a discreet, ear-worn EEG wearable that tracks brain activity day and night and uses AI to help people manage stress and sleep. Its research platform, AWEAR Labs, is used at Stanford. The company, led by founder and CEO Antonio Forenza, won Best Health Showcase Stage Pitch at TechCrunch Disrupt 2025 and is preparing its seed round.

YOU(th) came second, just two votes behind on Most Deployable. Led by Dr. Francesca Giovanetti, it says its goal is to bring preventive health check-ups to 8 billion people through a smartphone. It's about to start raising its Series A, with growth driven by clinic and pharmacy partnerships.

Aqua Theon and Cosmica tied for third.

Aqua Theon makes functional nutrition products for longevity. Cosmica Biosciences uses space biology, studying how human cells change in microgravity, to build tools that measure and treat aging.

The results lined up with the panel. The two top-voted companies both measure something real about a person and point to an action, rather than handing back a score.

What attendees said

Guests rated the night 9.0 out of 10 in a post-event survey of 31 attendees. Two in three gave it a 9 or 10.

What stood out most was the people on stage. Twenty of the 31 named the speakers as a highlight.

"I really appreciated the selection of high calibre speakers, such as a Stanford Professor," said Marlene Ronstedt of Fountain.Clinic. "I think it's important - especially in longevity - to hear from experts and not just some random influencers."

For others, the payoff came after the panel. "The panelists and keynote were absolutely excellent! Made for great networking conversations! Made some connections and already scheduled follow on meetings!" said Baljit Ubhi, a fractional executive.

Michael Gorham, CEO of Bon5 AI, was "very impressed with the Superpower team." Fareeha, founder of Forme, called it a "High quality panel and discussion." Odin, founder of Ambit Exchange, summed the night up in three words: "Warm, organized, informative."

The room also wants back in. Eighteen respondents said they'd like to speak at a future AI Insiders event, and four asked about sponsoring one.