Apheris
Apheris is a Berlin-based federated computing platform founded in 2019 by Robin Röhm (CEO, trained in medicine and banking at UBS) and Michael Höh (CTO, PhD in physics and computer science from BCG).
Profile
Apheris provides federated computing platforms that let pharmaceutical companies train AI models on pooled proprietary data without sharing the underlying data.
Apheris is a Berlin-based federated computing platform founded in 2019 by Robin Röhm (CEO, trained in medicine and banking at UBS) and Michael Höh (CTO, PhD in physics and computer science from BCG). The company enables pharmaceutical and life sciences organizations to build secure collaborative data networks, allowing companies to pool sensitive data for AI model training without centralizing proprietary information. The core innovation is Apheris' Compute Gateway, which keeps data on each organization's servers while enabling distributed model training through federated learning architecture.
The company pivoted in 2023 to focus exclusively on pharma and life sciences after early-stage work on general federated learning frameworks. That refocus proved decisive: revenue quadrupled between Q4 2023 and Q1 2024. By October 2025, Apheris had powered the AI Structural Biology (AISB) Network to train OpenFold3 across proprietary datasets from five major pharma companies—AbbVie, Johnson & Johnson, Astex Pharmaceuticals, Bristol Myers Squibb, and Takeda—in under ten weeks without any data centralization.
The company closed a €20.1 million Series A in January 2025 led by OTB Ventures and eCAPITAL, with existing backers Octopus Ventures and Heal Capital participating. As of June 2026, Apheris has ~50 employees and operates three core federated networks: AISB for structural biology, ADMET for drug metabolism and toxicity, and a Real-World Evidence network. Named customers include Roche, Johnson & Johnson, Sanofi, AbbVie, and AstraZeneca. The company is positioned at the intersection of AI infrastructure and regulatory data governance—a market gap created by pharma's inability to pool datasets due to competition and IP concerns.
Who buys this
- Large pharmaceutical and biotech companies building drug discovery AI models on internal structural biology data
- Healthcare systems and hospitals seeking to collaboratively train predictive models on patient data while maintaining privacy compliance
- Contract research organizations (CROs) orchestrating multi-party analytics across distributed clinical trial data
- Manufacturing and industrial companies testing federated approaches to quality control and process optimization across supply chain partners
- Diagnostics and research tool makers building foundational AI models trained on diverse proprietary lab datasets
Publicly disclosed clients
- AbbVie
- AstraZeneca
- Bristol Myers Squibb
- Genentech
- Johnson & Johnson
- Roche
- Sanofi
- Takeda
Strengths and what to watch
Strengths
- Proven product-market fit in pharma: revenue grew 4x post-pivot (Q4 2023–Q1 2024); eight major pharma companies now use the platform for active AI model training
- Federated learning technology that genuinely solves data sharing at scale: five competitors trained OpenFold3 on proprietary data simultaneously without centralization, a technical milestone in early 2026
- Strong tier-1 investor backing (OTB Ventures, eCAPITAL, Octopus Ventures) and scientific advisory board including OpenFold creator Mohammed AlQuraishi and drug discovery leaders from Exscientia and Isomorphic Labs
Watch for
- Narrow customer concentration in pharma and life sciences; limited revenue diversification beyond drug discovery and biomarker discovery workloads
- Increasing competitive pressure from cloud vendors (AWS, Azure, Google Cloud) shipping federated learning capabilities natively; risk of commoditization if Apheris becomes an API wrapper
- Regulatory and data governance uncertainty: pharma may revert to internal-only data use if federated learning transparency requirements tighten under privacy regulations (GDPR, HIPAA, upcoming EU AI Act synthesis rules)
Recent moves
Key Information
- Industry
- AI Safety & Security
- Founded
- 2019
- Headquarters
- Berlin, Germany
Frequently Asked Questions
What is Apheris?
Apheris is a Berlin-based federated computing platform founded in 2019 enabling pharmaceutical companies to train AI models on pooled proprietary data without centralizing datasets. Its Compute Gateway keeps data on each organization's servers while enabling distributed model training across networks.
How does federated learning work in pharma?
Federated learning allows multiple pharmaceutical companies to collaboratively train AI models while keeping proprietary data on their own servers. Instead of centralizing sensitive information, model training happens across distributed networks, eliminating IP and competition risks while preserving data privacy and compliance.
What is the AISB Network?
The AI Structural Biology (AISB) Network is Apheris's core platform enabling pharma companies to collaborate on protein prediction models. In October 2025, it trained OpenFold3 across proprietary datasets from five major pharma companies—AbbVie, Johnson & Johnson, Astex, Bristol Myers Squibb, and Takeda—without data centralization.
What companies use Apheris?
Named customers include Roche, Johnson & Johnson, Sanofi, AbbVie, AstraZeneca, Genentech, Bristol Myers Squibb, and Takeda. These eight major pharma companies use Apheris's federated networks for AI model training in drug discovery, structural biology, and drug metabolism research on proprietary datasets.
How much funding did Apheris raise?
Apheris closed a €20.1 million Series A in January 2025 led by OTB Ventures and eCAPITAL, with existing backers Octopus Ventures and Heal Capital participating. The funding followed the company's pivot to pharma focus, which quadrupled revenue between Q4 2023 and Q1 2024.
What products does Apheris offer?
Apheris operates three core federated networks: AISB for structural biology, ADMET for drug metabolism and toxicity, and a Real-World Evidence network. The Compute Gateway technology enables distributed AI model training while keeping data on each organization's servers without centralization or IP exposure.
How Apheris compares
Direct head-to-head against 2 competitors. Picked by 7wData.
Apheris
- Positioning
- Apheris provides federated computing platforms that let pharmaceutical companies train AI models on pooled proprietary data without sharing the underlying data.
- Customer segments
- Large pharmaceutical and biotech companies building drug discovery AI models on internal structural biology data
- Strengths
- Proven product-market fit in pharma: revenue grew 4x post-pivot (Q4 2023–Q1 2024); eight major pharma companies now use the platform for active AI model training
- Watch for
- Narrow customer concentration in pharma and life sciences; limited revenue diversification beyond drug discovery and biomarker discovery workloads
- Recent moves
- AISB Network Expands Federated OpenFold3 Initiative: Astex, Bristol Myers Squibb, and Takeda Join AbbVie and J&J to Fine-Tune Protein Prediction Models on Proprietary Data
Owkin
- Positioning
- Federated learning platform for pharma that also co-develops AI drug candidates. Positions as both infrastructure provider and drug discovery partner.
- Customer segments
- Large pharma companies (Sanofi, Novartis) and hospital networks. Buyers are heads of drug discovery or VP of data science.
- Strengths
- First commercial federated learning platform to secure strategic equity from a tier-1 pharma (Sanofi, $180M) while maintaining independent platform licensing.
- Watch for
- Co-developing drug candidates in same therapeutic areas as platform customers creates conflict. Pharma partners may restrict data sharing to protect competitive IP.
- Recent moves
- Launched K Pro agentic AI copilot for biopharma, October 2025. K Navigator copilot for biomedical research launched May 2025.
Rhino Federated Computing
- Positioning
- Federated computing platform for pharma and regulated industries. Operates the FAITE Biopharma Consortium across five active major pharma members.
- Customer segments
- Large pharma (AbbVie, Amgen, AstraZeneca, J&J, UCB via FAITE Consortium), hospitals, and clinical research networks. AI and data leads.
- Strengths
- FAITE Biopharma Consortium with five active pharma members (AbbVie, Amgen, AstraZeneca, J&J, UCB) running documented pre-competitive drug discovery workloads.
- Watch for
- $30M total funding to date limits capacity for multi-year enterprise pharma contracts. Pharma buyers flag vendor durability and long-term support in RFPs.
- Recent moves
- Raised $15M Series A led by AlleyCorp, May 2025. Rebranded from Rhino Health to Rhino Federated Computing, January 2025.
Sources
- techcrunch.com — Series A funding round details, product pivot in 2023, revenue growth metrics (4x), customer names (Roche, hospitals), CEO statement on data ownership
- www.apheris.com — AISB Network expansion October 2025, new member companies (Astex, BMS, Takeda), OpenFold3 training achievement, federated learning technology details
- www.apheris.com — Core product offering, AISB Network members, ADMET Network, ApherisFold, customer testimonials, company website overview
- www.eu-startups.com — Series A funding announcement January 2025, OTB Ventures and eCAPITAL as lead investors, strategic direction on life sciences data networks
- www.apheris.com — Founders Robin Röhm and Michael Höh, scientific advisory board members, company mission and values, office locations
- pharmaphorum.com — OpenFold3 consortium expansion details, partner statements from BMS and Astex, federated learning approach and impact
- www.glassdoor.com — Employee feedback on company culture, team quality, mission impact, work-life balance ratings
- discover-pharma.com — AISB Network expansion with new members Astex, BMS, Takeda, federated learning security benefits, Columbia University partnership