Apheris: Powering global life sciences data networks

We are delighted to announce that we have led the Series A round in Apheris, an emerging leader in federated AI infrastructure, alongside eCAPITAL, Octopus Ventures, Heal Capital and others. We look forward to supporting Apheris in its mission to power global data networks in life sciences and healthcare. 

“Decentralized AI” theme 

Currently, the prevailing status quo is to execute AI workloads - such as training, inference, data processing, and analytics - in centralized locations. At OTB, we hypothesize that a substantial portion of AI computing will need to shift to a local, decentralized approach. 

We see several reasons for decentralized AI architectures: 

• Data fragmentation: Public data is vastly overshadowed by proprietary private data, which remains siloed across numerous market participants. Foundation models have already reached the limits of absorbing data from public domain and require additional input. 

• Data privacy: Concerns about the misuse of proprietary, sensitive and personally identifiable data are increasing. Several regulations have been introduced (GDPR, EU AI Act), and we anticipate that this trend will continue. 

• Data routing costs:  Moving petabytes of raw data to the centralized servers is costly. This issue will be exacerbated by growing AI adoption outside Big Tech and rise in edge computing. 
   

Federated computing paradigm in life sciences and healthcare 

The classic centralized approach to AI/ML is particularly problematic in pharma and healthcare. Data is distributed across various stakeholders (pharma, biotech, hospitals, insurers) who cannot share it due to regulatory and commercial sensitivities. As a result, it is estimated that a staggering 97% of all healthcare data remains completely unused. 

Federated architecture might be complex, but its core philosophy is simple - instead of transferring sensitive raw data to a centralized model, computations are executed locally where data resides, and only the outputs (e.g., model parameters) are aggregated centrally. This technology was initially adopted by Big Tech discretely in their own products. We now see a robust ecosystem of third-party and open-source software tools that could unlock broader market adoption. 

Apheris on a mission to power global life sciences data networks 

Apheris is at the heart of a federated AI setup. Its product is a lightweight software agent which serves as a secure gateway between local data and central AI models. It safeguards sensitive data and empowers data owners to control which computations are performed. The product also enables seamless integration with complementary privacy-enhancing technologies (homomorphic encryption, differential privacy, synthetic data). Its neutral middleware positioning fosters trust among stakeholders, unlocking the potential for cross-institutional collaboration. 

Apheris is both use-case and AI-model agnostic, offering a diverse range applications in drug discovery, real-world evidence studies, and clinical trial design. Over time, its product could also potentially be used in other industries, such as finance, defense or manufacturing. 

Strong initial traction and future ambition 

Within the broader data privacy landscape, Apheris stands out for its ability to deliver a scalable enterprise-grade product. This is evidenced by several contracts already signed with several of the world’s top 20 pharmaceutical corporations, who have deployed Apheris in their live production environments. While still early in its journey, Apheris could become a critical component in world’s leading federated data networks, leveraging strong network effects over time. 

The company plans to use its Series A funding to host leading foundation models, such as those for protein-complex-prediction. These models will aim to surpass state-of-the-art performance by utilizing unprecedented access to more training data that is currently locked in private silos. 

Team 

Last but not least, we are impressed with people behind Apheris - Robin, Michael and the team they have assembled. Throughout our due diligence interviews with clients, investors, and other ecosystem participants, one consistent message stood out — while Apheris’ value proposition is compelling on its own, it is the team behind it that ultimately moves the needle.  
 
Marcin Hejka and Mateusz Lukasik from OTB will join Apheris’ advisory board. We are thrilled to support Robin and Michael on their exciting yet demanding journey. 

You can read more about Apheris here:

Tech Crunch

Medium