Calico and the AI-Driven Search for the Molecular Mechanisms of Aging

1. Introduction: Calico and the Bet on Deep Aging Biology

Aging is the single largest risk factor for most chronic disease, yet it is not itself a regulatory indication. This mismatch frames the central strategic problem of geroscience: the biology of aging can be studied and, in model organisms, modified, but the resulting knowledge must be converted into therapies for conventional diseases — neurodegeneration, fibrosis, cardiometabolic disease, cancer — before it can reach patients. Calico Life Sciences (the California Life Company) was created in 2013 as an explicit wager that patient, well-funded basic biology is the correct route through that bottleneck.

The company was founded by Alphabet (then Google) and Arthur D. Levinson, the former Genentech chief executive and chairman of Apple, who has led it since inception. Calico is headquartered in South San Francisco and, from the outset, described itself as focused on the biology of aging and age-related disease rather than on consumer longevity products. That positioning — a research organisation with a pharmaceutical horizon, financed by a technology company — distinguishes it from both classical biotech and the consumer-facing longevity industry.

A decade on, the record is mixed and instructive. Calico has built genuine scientific capability and a clinical pipeline, but in 2025 it suffered a high-profile trial failure and the termination of its founding pharmaceutical partnership. This article separates what is established, emerging, and merely claimed about Calico’s model, its science, and its computational approach — and uses that record to examine why translating aging biology is so difficult.

2. Origins, Funding, and Organisational Model

Calico’s founding capital came from Alphabet, which framed the venture as a long-horizon bet rather than a near-term product line. Levinson’s recruitment signalled that the intent was drug development, not publishing alone: he had run Genentech through the era in which recombinant biology became medicine. The company’s stated mission — to understand the biology of aging and use that understanding to develop interventions for age-related diseases — has remained stable, even as its partnerships have changed.

The company’s most consequential early decision was a 2014 research collaboration with AbbVie. Under the original terms reported at the time, each partner committed an initial sum on the order of a quarter of a billion dollars, and the agreement was extended in 2018 with further commitments from both sides. According to AbbVie’s 2024 annual report, the pharmaceutical company invested approximately $1.75 billion into the collaboration between 2013 and 2022. This is an established, reported figure, not an estimate. The structure was unusual: a young, discovery-focused organisation partnered with a large pharmaceutical company for shared risk and downstream development capacity.

The organisational bet was that depth and duration would outperform speed. That bet remains under evaluation: as of late 2025 the collaboration had produced a small number of clinical-stage assets but no approved drug.

3. Research Strategy: Comparative Biology, Genetics, and Longitudinal Phenotyping

Calico’s public research programme spans several traditions in aging biology. They differ substantially in how strong their translational foundation is, and it is useful to rank them accordingly.

3.1. Comparative Biology of Long-Lived Species

Some mammals live far longer than their body size would predict — the naked mole-rat being the standard example, with a reported lifespan exceeding thirty years and unusual resistance to cancer. Studying such species can surface protective mechanisms, but the evidence is largely correlational: a mechanism that correlates with longevity in one species does not establish that it causes longevity, and it does not establish that it can be safely transferred to humans. This work is best classified as hypothesis-generating.

3.2. Genetics of Lifespan in Model Organisms

Lifespan genetics was built on yeast, worms, flies, and mice. Over decades this work established that a small number of conserved nutrient-sensing pathways — mechanistic target of rapamycin (mTOR), insulin/IGF-1 signalling, AMP-activated protein kinase (AMPK), and sirtuins — can be modulated to extend lifespan in these organisms. This is established in model systems. The important caveat is that effect sizes and mechanisms differ across species and even across mouse strains, and no intervention has yet been shown to extend maximum human lifespan. The geroscience framework that Calico operates within — the idea that aging biology itself can be a therapeutic target for multiple diseases at once — is set out in the literature and supported by genetics, but its clinical realisation remains emerging.

3.3. Deep Phenotyping and Aging Clocks

A third strand is systematic measurement: collecting large, longitudinal datasets — molecular, physiological, imaging — so that the trajectory of aging can be modelled rather than inferred from single time points. Aging clocks trained on such data can predict mortality or morbidity at the cohort level. Whether clock readings are causal mediators that can be therapeutically modified, or downstream markers that merely track accumulated damage, is unresolved. This distinction matters because a treatment that changes a clock reading has not thereby been shown to change health outcomes.

4. Where AI Enters: Computation as a Discovery Engine

Calico has publicly emphasised computational and machine-learning methods as central to its strategy: analysing large omics and imaging datasets, triaging targets, and modelling the interaction between small molecules and aging-related proteins. The general premise is defensible. Machine learning is now a standard tool in target identification, in analysis of high-dimensional biological data, and in compound property prediction.

Two clarifications are essential to avoid the hype that surrounds «AI-driven aging research.»

• Structural prediction is not drug discovery. The arrival of accurate protein structure-prediction tools transformed structural biology and is genuinely established as a research capability. But predicting a structure does not identify a tolerable drug, a safe dose, or a clinical benefit; those remain wet-lab and clinical problems.
• «AI-found target» is not a validated target. A model can rank hypotheses far faster than humans, but a ranked hypothesis still requires experimental and, ultimately, clinical validation. Public, peer-reviewed demonstrations that an AI-originated aging target reached late-stage clinical success are, at present, essentially absent across the field — a limitation to state plainly rather than obscure.

The most accurate description of Calico’s contribution here is therefore emerging: computational methods shorten parts of the discovery cycle and expand the space of reachable hypotheses, while the causal and clinical questions remain to be settled experimentally. Claims of a general acceleration of longevity drug discovery through AI are, for now, forward-looking company framing rather than demonstrated outcome.

5. The Clinical Pipeline: From eIF2B to Anti-PAPP-A

Calico’s programmes moved into human testing through the AbbVie collaboration. The company has stated that five assets advanced into clinical development in neuroscience and oncology, with more than twenty early-stage programmes behind them — figures that originate from Calico and should be treated as company-reported. Two assets illustrate the range.

5.1. Fosigotifator and the Integrated Stress Response

Fosigotifator (ABBV-CLS-7262) is a small-molecule activator of eukaryotic initiation factor 2B (eIF2B), a control node in the integrated stress response. The rationale is that chronic over-activation of this pathway impairs protein synthesis and contributes to the accumulation of misfolded or aggregated proteins in neurodegenerative disease. This is a plausible, mechanism-based hypothesis rather than a validated human mechanism.

The drug was tested in amyotrophic lateral sclerosis (ALS) as one regimen of the HEALEY ALS Platform Trial, a multi-arm adaptive platform run at Massachusetts General Hospital. In January 2025, Calico and the trial investigators reported that at the primary dose the regimen did not meet the primary endpoint of disease progression (a combined analysis of function and mortality) after 24 weeks, and that key secondary endpoints were not statistically different from placebo. An exploratory high-dose group showed a possible signal on muscle strength measured by hand-held dynamometry; such exploratory findings, in a non-primary comparison, are hypothesis-generating and cannot be read as efficacy. Fosigotifator was also being developed for vanishing white matter disease and major depressive disorder, testing different scientific hypotheses.

5.2. ABBV-CLS-628 and PAPP-A

A second clinical asset, ABBV-CLS-628, is an antibody against pregnancy-associated plasma protein A (PAPP-A) under study for autosomal dominant polycystic kidney disease. It reportedly entered mid-stage clinical testing and received a regulatory fast-track designation in 2025. Here the biology is more directly tied to a defined disease process than to «aging» as such — a reminder that most geroscience reaches patients through named indications.

The pattern across the pipeline should be read carefully. Targeting fundamental aging biology does not remove the ordinary difficulty of drug development: most candidates fail, and a failure does not by itself refute the underlying biology. The fosigotifator result is best described as an informative negative on one target and one dose range, not as a verdict on geroscience.

6. The AbbVie Split and What It Signals (2025-2026)

In November 2025, AbbVie ended the partnership after about eleven years. Reporting based on internal communications indicated that roughly 100 staff associated with the collaboration would be laid off, and that AbbVie was reorienting investment toward biologics, cell therapy, and RNA-based medicines rather than small molecules. A Calico spokesperson confirmed the end of the collaboration and stated that Calico is resourced to continue advancing the programmes independently while seeking new partnerships. Separately, AbbVie disclosed a gain associated with the termination of the collaboration in its 2025 accounts.

Calico also created a new drug-discovery leadership role, appointing Philip Kym, previously a senior medicinal chemist at AbbVie, to head discovery in October 2025. Two readings of the split are possible, and both are supported by the facts:

• A strategic and commercial decision by AbbVie. After more than a decade and a large investment without an approved product, a portfolio pivot away from small molecules is a rational corporate choice that need not reflect poorly on the science.
• A translation problem in the aging field. The episode illustrates how hard it is to convert aging biology into approved medicines on the timescale a large pharmaceutical partner requires.

What it does not demonstrate is that Calico’s scientific premise is false, nor that aging is not druggable. It demonstrates the cost and duration of the attempt.

7. The Translation Problem: Why Aging Is Hard to Drug

Calico’s experience is a case study in a structural difficulty. Several features of aging biology conspire against rapid clinical success:

• No accepted indication. Regulators approve drugs for diseases, not for aging. A programme must therefore pick a disease endpoint, which may be only partly sensitive to the aging mechanism being targeted.
• Long timelines. Interventions intended to slow degeneration require lengthy trials and large samples, particularly when the effect on any single endpoint is modest.
• Pleiotropy and safety. Pathways such as mTOR and eIF2B are central to normal physiology. The more fundamental the target, the greater the risk of on-target toxicity, as the side-effect profile of several geroprotective candidates has shown.
• Surrogate endpoints. Aging clocks and other biomarkers are surrogates. Regulatory bodies generally require a surrogate to be validated as predicting clinical benefit before it can substitute for hard outcomes such as disease events or mortality — and most aging biomarkers do not yet meet that bar.
• Heterogeneity. Individuals age at different rates and through different mechanisms, so a single mechanism-based drug may benefit only a subgroup, diluting apparent efficacy in an unselected trial.

A responsible assessment of Calico is inseparable from this context. The field’s critics and its advocates agree on the obstacles; they disagree on how long patience is warranted.

8. What Calico Has and Has Not Demonstrated

The honest summary is that Calico has demonstrated that large-scale, patient aging research can be organised and sustained, and that it can generate clinical candidates — but not yet that it can deliver a therapy that changes the course of an age-related disease. The datasets and methods it has built may prove valuable to the wider field, but that value is prospective rather than established.

9. Conclusion

Calico represents the most sustained attempt yet to make aging biology a pharmaceutical discipline, combining comparative biology, lifespan genetics, deep phenotyping, and computation under one roof. It has produced clinical candidates but no approved drug, and its founding partnership ended in 2025 on the back of a trial failure and a shift in its partner’s strategy. The correct conclusion is neither that the enterprise failed nor that it is on the verge of success, but that it has clarified the size of the problem: target validation in aging is slower, costlier, and less forgiving than in conventional drug discovery, and computational tools shorten the path without removing the clinical bottleneck.

Calico’s ultimate contribution will be judged not by press releases or by the sophistication of its models, but by whether any molecule it advanced produces a measurable clinical benefit in a named disease. Until then, the appropriate posture toward claims of AI-accelerated rejuvenation — from Calico or anyone else — is calibrated interest, with the burden of proof resting on published, independently replicated outcomes.

Sources and Key Research Centers

1. Calico Life Sciences (California Life Company) — Company publications, partnerships, pipeline, and scientific strategy.
2. Calico: fosigotifator in the HEALEY ALS Platform Trial (January 2025) — Topline results for Regimen F, including the primary-endpoint failure at the primary dose.
3. ClinicalTrials.gov NCT05740813: HEALEY ALS Platform Trial — Regimen F (ABBV-CLS-7262) — Registered trial record for fosigotifator.
4. Kennedy BK et al. «Geroscience: linking aging to chronic disease.» Cell, 2014; 159(4): 709-713. DOI: 10.1016/j.cell.2014.10.039 — Framework for treating aging biology as a target for multiple diseases.
5. López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G. «Hallmarks of aging: An expanding universe.» Cell, 2023; 186(2): 243-278. DOI: 10.1016/j.cell.2022.11.001 — Current consensus map of aging mechanisms referenced by the field.
6. PubMed: Calico-affiliated publications — Peer-reviewed output on lifespan genetics, comparative biology, and aging biomarkers.
7. National Institute on Aging (NIA) — Public-sector counterpart in aging biology and target discovery; source of the geroscience framework.
8. Endpoints News: AbbVie and Calico end 11-year partnership (November 2025) — Reporting on the termination of the collaboration.
9. Nature Reviews Drug Discovery: industry-analysis commentary on Calico’s move to independence (2025) — Trade-press analysis of Calico’s transition.
10. AbbVie SEC filings (Form 10-K) — Disclosure of cumulative investment in the Calico collaboration and of the termination-related gain.