apheris AI

Superior drug discovery models through federated data networks

Federated AI and Secure Data Collaboration Growth Series A

Overview

Apheris AI is a Berlin-based deep tech company that builds federated computing infrastructure enabling life sciences organizations to collaborate on sensitive, proprietary data without centralizing it. By keeping data within each organization's own secure environment while enabling joint AI model training and analytics, Apheris addresses one of the most significant barriers to AI in drug discovery: the inability to share patient data, genomic information, and compound libraries across institutional boundaries.

The company's core technology -- federated learning -- allows AI models to travel to the data rather than requiring data to travel to the models. This approach enables pharmaceutical companies, hospitals, and research institutions to pool their AI insights while each participant retains full control of their proprietary datasets. Apheris has found strong product-market fit in life sciences, powering multiple cross-institutional data networks including the AI Structural Biology (AISB) Network, the ADMET Network, and the Antibody Developability Network. The company reports multiplying its revenue by four times since launching its current product in late 2023.

In January 2025, Apheris raised a EUR 20.1 million Series A round co-led by OTB Ventures and eCAPITAL, with participation from existing investors Octopus Ventures and Heal Capital. While the company primarily serves pharmaceutical R&D, its federated data collaboration technology has documented applications in financial services and insurance, where carriers can evaluate external data and models under strict infosec and legal constraints without exposing proprietary datasets.

Products & Services

Apheris Gateway

Core federated computing platform enabling organizations to query, analyze, and train AI models on distributed data without moving or exposing underlying datasets. The Gateway is deployed within each participant's own secure infrastructure, allowing model training and analytics without data egress.

Key Features

Secure local inference so data, queries, and outputs remain in-house
Fine-tuning capabilities for proprietary datasets using parameter-efficient methods (LoRA and FRA-LoRA)
Benchmarking against internal datasets
GUI and API access to industry-trained models

Target Users: Pharmaceutical companies, research institutions, and enterprise data teams requiring compliant cross-institutional AI collaboration

ApherisFold

Enterprise application for protein structure prediction using co-folding models deployed securely within a company's own IT environment. Supports OpenFold3, Boltz-2, and Protenix-v1. A free Lite version is available for browser-based exploration without infrastructure setup.

Key Features

Run, benchmark, and fine-tune state-of-the-art co-folding models (OpenFold3, Boltz-2, Protenix-v1)
Monomer and multi-chain structure prediction, protein-ligand complex modeling, and binding affinity estimation
Interactive 3D visualization for scientific review
API access for integration into existing drug discovery pipelines
Secure local execution -- all data remains within enterprise environment
Auditable, reproducible records for regulated environments

Target Users: Computational chemists, structural bioinformaticians, and medicinal chemists in pharmaceutical and biotech organizations

Federated Data Networks

Industry-specific collaborative networks powered by the Apheris platform, enabling multiple organizations to jointly train AI models on pooled data without exposing proprietary datasets.

Key Features

Pre-competitive data sharing without IP exposure
Network-wide model improvement as more organizations participate
Governance framework for data access and usage

Target Users: Pharmaceutical companies, biotech firms, and research consortia seeking collaborative AI model development

At a Glance

Founded: 2019
Headquarters: Berlin, Germany
Employees: 51-200
Funding: Series A

Category & Focus

Category: Federated AI and Secure Data Collaboration
Subcategories: Federated learning privacy-preserving machine learning cross-institutional AI model training
Insurance Verticals: P&C Life & Annuity (secondary/adjacent use cases)
Target Customers: Carriers, Solution Providers