Unlocking the power of consumer genetics: 15 million genomes at Regeneron's fingertips

What happened

Regeneron acquired nearly all of 23andMe's genetic database and research assets for $256mn during a bankruptcy auction

In a court-supervised bankruptcy auction, Regeneron Pharmaceuticals Inc agreed to purchase the majority of 23andMe's assets, excluding its telehealth unit, Lemonaid Health. The acquisition encompasses 23andMe's Personal Genome Service, Total Health and Research Services, and a biobank containing genetic data from over 15mn customers. 

The deal, pending regulatory and bankruptcy court approvals, is expected to close in the third quarter of 2025. 

Impact on our Investment Case

The data is vast and, best of all, cheap

Of course the dataset is not just 15mn genomes, as in a 15mn word documents with only a string of ATCG letters. The data comes with user profiles, surveys, family information, contact information, and more, like DNA-based social network data allowing users to connect with genetic relatives and share their ancestry and health information. The metadata is far more valuable than the genome itself.

Not only this but the majority of these people are still alive and they are not patients. Additionally, ~80% of the accounts were consented and opted-in for research. That is $17 per account and $21.3 per opted-in for research genome with huge amount of metadata. Even comparing to the $200 for genome sequencing by Illumina's NovaSeq X series, it is very cheap.

The data set is bigger and cheaper than the last deal

This data set is significantly larger than Amgen's deCODE, for which Amgen paid $415mn. It had 140k genotyped individuals, and only really 2.6k fully sequenced genomes when bought in 2012. This represented closer to $2900 per genome, bearing in mind that sequencing a single genome around 2010 costed ~$19500. As you may remember, deCODE also went bankrupt.

The 23andMe data set is a bit messier and carries more risk of lawsuits than the deCODE database. For EU clients, GDPR rules could be used to remove data. Nonetheless, the scale and breadth of the data are worth the risk, and Regeneron, unlike 23andMe in 2023, has never suffered a data breach.

The conundrum of genomic data business is that it is challenging to monetize while at the same time being very valuable. 23andMe's founder Wojcicki had even plan to take the company private after its stock value collapsed by more than 97% compared to its multibillion-dollar market debut just three years ago. But everything was falling apart: the data breach event, independent directors resigning from the board, declining revenue in testing, clinical pipeline slashed... 

Regeneron and 23andMe already had a deal

In 2013, 23andMe received a FDA warning letter that temporarily halted its health-related genetic reporting in the U.S. due to concerns over the accuracy and clinical validity of its tests.

In 2014, Regeneron established the Regeneron Genetics Center to sequence and analyze large-scale genetic data. 

By 2015, 23andMe was rebounding, regaining FDA approval and expanding its research ambitions, and Regeneron was keen on boosting its sequencing effort. Thus they signed a 5-year deal, with potential extensions, for which the specific financial terms were not publicly disclosed.

Regeneron gained access to de-identified genetic data from 23andMe’s customers who consented to participate in research, while 23andMe received money. So far no specific drugs directly attributed to this deal have been publicly highlighted.

AI is the key driver of the deal

Now Regeneron has all the data, and the means to extract far more value from it than it could in 2015

Key Differences:

Aspect	2015	2025
Data Analysis	Relied on statistical methods and early bioinformatics; limited to simple patterns.	AI (deep learning, LLMs) detects complex genetic interactions across 15mn profiles.
Drug Target Discovery	Manual, hypothesis-driven; slow validation via wet-lab experiments.	AI predicts causal variants, simulates drug interactions, and prioritizes targets.
Personalized Medicine	Limited to population-level studies; minimal patient stratification.	AI enables patient subgrouping and trial matching for targeted therapies.
Computational Efficiency	Costly, slow analysis due to limited cloud and AI capabilities.	Cloud, GPUs, and automated AI pipelines reduce costs and accelerate insights.
Data Integration	Siloed data; minimal external linkage due to privacy and technical limits.	AI integrates with health records, real-world evidence via federated learning.
Regulatory/Ethical	Evolving regulations; limited AI frameworks slowed adoption.	AI aligns with FDA guidelines; privacy-preserving techniques ensure compliance.

This acquisition could meaningfully enhance Regeneron's ongoing R&D efforts, particularly in oncology. By leveraging genetic and lifestyle data, the company can better identify mutations driving resistance to current therapies and refine patient stratification for clinical trials, ultimately improving the design and success rate of precision cancer treatments.

More importantly, it opens the door to entirely new opportunities. With such a rich dataset, Regeneron could broaden its focus beyond oncology into preventive medicine, such developing polygenic risk scores for early intervention or uncovering multi-indication therapies through shared biological pathways. Polygenic risk scores are personalized estimate of how likely you are to develop a particular disease based on your genes.

That said, the payoff won't be immediate. This is a long-term play likely spanning a decade. Still, for a stock in decline since Q4 last year, this deal adds a compelling strategic angle. 

Our Takeaway

Yes, Regeneron was always a frontrunner to acquire 23andMe’s assets, given their existing partnership. But the price paid and the scale of the dataset make this one of the most impactful genomics data deals of the decade.

At atonra, we currently have no position in Regeneron, but once the Eylea overhang clears and market focus returns to pipeline potential, current levels represent an attractive entry point. In the meantime, we are already exposed to companies deeply embedded in genomic data. This acquisition reinforces our conviction that Big Pharma increasingly recognizes the strategic value of large-scale, high-quality genetic datasets.