Deerfield, which sold startup Nuvalent to GSK for $10,6 billion, calls “longevity” a misnomer and is building a New York center to test longevity claims against evidence

Deerfield, which sold startup Nuvalent to GSK for $10,6 billion, calls “longevity” a misnomer and is building a New York center to test longevity claims against evidence

On 6 October 2026, Jim Flynn, managing partner and CEO of healthcare investment firm Deerfield Management, told Longevity.Technology on the sidelines of ARDD 2026 that his firm is building a nonprofit longevity center in New York. He says there is no reliable way to measure the claimed extension of healthy life, which is why he considers “longevity” a misnomer for the category itself. The center will take one and a half to two years to establish.

Flynn has worked in healthcare investing for 38 years. Deerfield, which he leads, was founded in 1994 with $17 million in capital and has grown into one of the world’s largest healthcare investors, with more than $16 billion under management as of December 2025. He describes his approach this way: an investor’s task is “to narrow the error bars around the numbers rather than find the exact numbers.” In 2015, geneticist Matt Nelson showed in Nature Genetics that a drug targeting a gene statistically associated with a disease is almost twice as likely to receive approval. Deerfield hired him, and the approach proved successful with Nuvalent: the firm founded the company, retained full ownership for five years, took it public in an $800 million IPO, and sold it to GSK four years later for $10,6 billion.

Flynn applies the same reasoning to “longevity” as a whole. Longevity biotech companies seek drugs that could extend healthy life, but evaluate them and seek approval for a different indication that is easier to measure. Flynn regards any longevity benefit as a bonus because there is still no measurable trial endpoint for life extension itself that regulators would accept. About five days earlier, at the same ARDD conference, the regulator itself had publicly grappled with this missing measure: during an FDA panel on trial design for aging interventions, aging clock developer Steve Horvath insisted from the stage, “We cannot get there without molecular biomarkers of aging.” Longevity.Technology estimates that an additional $173 billion in annual sales is at stake, but only if a dedicated approval pathway for anti-aging drugs becomes available. For Flynn, gene therapy offers a “straight line”: it corrects the absence of a single protein. Aging involves too many factors to attribute to a single cause, making the case for a drug targeting it “murky.”

Deerfield’s response is an institute intended to provide an impartial reference point in a market where almost everyone is selling something. A free digital service analyzes diagnostic data and provides recommendations without a commercial interest; a paid, in-person service provides access to procedures that Deerfield tests in its own laboratory inside the Cure building, where the firm has around 30 university partnerships.

“None of this is for profit. It is all about providing truly neutral information, largely because I am simply very tired of all this misinformation,” Flynn says.

Flynn applies the same standard to AI for drug development:

“ChatGPT worked because it used words rather than letters. In chemistry, it is very difficult to apply the same approach to atoms,” Flynn says.

A language model learns quickly because each predicted word can be checked immediately against the text. A chemistry model has no equivalent check: the molecule it predicts must be synthesized and tested in a laboratory, a cycle that takes three months rather than moments. Deerfield’s response is to synthesize and test large numbers of molecules at high throughput to achieve a similarly rapid learning cycle.

Flynn predicts that the next breakthrough in clinical trials will come from digital disease models that the FDA accepts as a new type of trial endpoint. Instead of lengthy assessments of Parkinson’s tremor using a clinician-rated scale, patients would be compared with an impartial digital model of their disease.