Retro Biosciences

Retro Biosciences is a privately held longevity company founded in 2021 and funded initially with $180 million from Sam Altman, with the stated objective of adding ten years to healthy human lifespan. It runs three programmes — small molecules that enhance Autophagy, cellular reprogramming, and therapies derived from the young-plasma literature — and is more specific in public than its better-funded peers about programmes, goals and timelines. An early collaboration with OpenAI on protein design has drawn most of the attention to it.

§Overview

The company's framing is deliberately narrower than the field's usual rhetoric. Rather than treating aging as a single process to be reversed, it selected three mechanisms with existing preclinical support and organized around getting each into humans. Joe Betts-LaCroix, a hardware engineer turned entrepreneur who previously co-founded a computing company and had been involved in longevity advocacy, is chief executive.

Retro is smaller than Altos Labs by an order of magnitude in capital, and its public posture is correspondingly more concrete: it names its programmes, states a numerical goal, and has discussed clinical timelines, none of which its larger competitor does.

§History

The company was formed in 2021 and operated quietly until March 2023, when MIT Technology Review reported that Sam Altman, then chief executive of OpenAI, had personally invested $180 million — the whole of the company's funding at that point.¹ Altman has described the investment as motivated by an interest in extending healthy life rather than by a commercial thesis, and holds no operational role.

In January 2025 the same publication reported a collaboration between Retro and OpenAI on a model called GPT-4b micro, trained to propose redesigned versions of proteins in the manner described under AI protein design.² The reported application was engineering variants of the Yamanaka factors — specifically SOX2 and KLF4 — to improve reprogramming efficiency, with Retro claiming a large increase in the expression of markers associated with pluripotency induction. The result was announced through press coverage rather than a peer-reviewed paper, and independent replication has not been published.

1. 2021
   Founded
   Retro Biosciences is established in the Bay Area with three programme areas and a stated ten-year healthspan goal.
2. 2023
   Funding disclosed
   MIT Technology Review reports Sam Altman's $180 million personal investment, at the time the company's entire capital.
3. 2025
   Protein design collaboration
   Retro and OpenAI report a purpose-built model proposing improved Yamanaka factor variants, announced via press rather than publication.
4. 2025–2026
   Toward the clinic
   The company moves its lead autophagy-enhancing compound toward first-in-human testing and reports seeking substantially larger funding.

§Research programme

§Autophagy enhancement

The most developed programme targets autophagy, the lysosomal degradation pathway whose molecular machinery Yoshinori Ohsumi worked out in yeast and for which he received the 2016 Nobel Prize in Physiology or Medicine.³ Autophagic flux declines with age in most tissues studied, and genetic or pharmacological enhancement extends lifespan in flies, worms and mice; nearly every intervention that reliably extends lifespan in model organisms, including Caloric restriction and Rapamycin, appears to require an intact autophagy pathway, which is why it recurs across the Hallmarks of aging as a shared downstream node. Retro's approach is a small molecule intended to raise flux without the immunosuppression that accompanies mTOR inhibition, with an initial indication in neurodegeneration, where failing Proteostasis collapse is a well-established feature of disease.

The gap between the model-organism evidence and human benefit is wide and the company acknowledges it. No autophagy-enhancing drug has demonstrated a clinical benefit in an aging-related indication in humans.

§Cellular reprogramming

The reprogramming programme pursues the same goal as Altos Labs and NewLimit: restoring youthful cell function without loss of cell identity. It differs from both in route. Retro works on the reprogramming factors themselves, engineering variants of the canonical set, which is where the OpenAI collaboration was applied, while NewLimit treats the canonical factors as the wrong tool and screens for combinations that never approach pluripotency. Neither route removes the delivery problem.

§Plasma-derived therapeutics

The third programme derives from heterochronic parabiosis, in which young and old mice share a circulation and the old animal shows improvements in several tissues. The interpretation of those experiments is contested — whether the effect comes from youthful factors added or from aged factors diluted — and the plasma-transfusion clinics that grew up around them drew a public warning from the US Food and Drug Administration in 2019. Retro's stated approach is to identify specific circulating factors rather than to transfuse plasma, which sidesteps the strongest objection to the commercial versions.

Mice, not people

Every mechanism Retro works on rests on rodent evidence. Autophagy enhancement, partial reprogramming and heterochronic blood exchange have all produced functional improvements in mice. None has been shown to extend healthy life in humans, and the company's ten-year target is a goal, not a projection supported by data.

§Funding and structure

Altman's $180 million was, at the time it was reported, one of the largest single-individual investments in a longevity company. Reports in 2025 indicated the company was raising a substantially larger round, and Retro has said it intends to fund clinical development rather than remain a discovery organization.

The funding structure raises a governance question the company has addressed only briefly. Altman was simultaneously the chief executive of OpenAI and the sole funder of a company that entered a collaboration with OpenAI, an arrangement that would attract scrutiny in a public company. Retro's position is that the collaboration was conducted on ordinary commercial terms.

§Reception

The scientific reception has been more favourable than that given to most well-publicized longevity ventures, largely because the company works on established mechanisms rather than novel claims and has been explicit about the preclinical state of the evidence. Specialists who are skeptical of the field's rhetoric have noted that autophagy enhancement is a reasonable target with a plausible path to an approvable indication.

Criticism has focused on three points. The ten-years-of-healthy-life goal has no operational definition; Healthspan is not a standardized measure, no qualified surrogate endpoint exists, and no trial design has been proposed that would demonstrate the claim within the lifetime of the company. The GPT-4b micro result was communicated through journalism rather than publication, which is the pattern the field has been criticized for since the resveratrol episode associated with David Sinclair. And a company organized around a specific numerical promise faces an incentive to define success downward, which the absence of a preregistered endpoint makes easy.

§Outlook

Retro's near-term significance is that it may be first among the well-funded reprogramming and geroprotection companies to generate human data. A phase-one safety result for an autophagy enhancer, followed by a phase-two efficacy signal in a neurodegenerative indication, would be the first time any of this generation of longevity companies has produced evidence a regulator recognizes. A null result would be equally informative, and considerably more likely on base rates for central nervous system drugs.

The AI-assisted protein engineering strand is the more consequential unknown. If purpose-trained models can reliably propose transcription-factor variants with improved function, that changes the cost structure of protein engineering across the field, well beyond Epigenetic reprogramming. The claim as reported is large enough that the absence of a peer-reviewed description more than a year later is itself a data point.

§See also

• Autophagy
• Altos Labs
• NewLimit
• Partial reprogramming
• Parabiosis and young blood
• Healthspan
• Geroscience hypothesis
• Rapamycin

§References