Apheris

Cover Block

PUBLIC

Name	Apheris
Tagline	Federated infrastructure for training AI on sensitive pharma data without centralizing it. [apheris.com]
Headquarters	Berlin, Germany
Founded	2019 [TechCrunch, Jan 2025]
Stage	Series A
Business Model	SaaS
Industry	Deeptech
Technology	AI / Machine Learning
Geography	Western Europe
Growth Profile	Venture Scale
Founding Team	Co-Founders (2)
Funding Label	Series A (total disclosed ~$20,800,000) [Crunchbase, Retrieved 2026]

Links

PUBLIC

• Website: https://www.apheris.com/
• LinkedIn: https://www.linkedin.com/company/apheris/
• X / Twitter: https://x.com/apheris_AI

Executive Summary

PUBLIC

Apheris provides the technical and governance infrastructure to train AI models across the highly sensitive, proprietary datasets of pharmaceutical companies without requiring that data to leave its original secure environment. This federated approach directly addresses a critical bottleneck in drug discovery, where valuable data is often siloed by regulation and competition, making the company a notable player in the emerging intersection of AI, life sciences, and data privacy [TechCrunch, Jan 2025]. The founding story is rooted in this problem, with CEO Robin Röhm having previously lost customers at a prior startup due to the inability to centralize data for analysis under regulatory constraints [apheris.com].

Its core product, the Apheris Gateway, is positioned as an end-to-end solution that enables collaborative machine learning, allowing competing or cooperating entities to build better predictive models while maintaining strict control over their data assets [apheris.com]. The founding team combines a deep understanding of the life sciences domain with technical execution: Röhm's background spans medicine, philosophy, and mathematics, while CTO Michael Höh holds a PhD in physics and computer science and previously built AI applications at BCG [apheris.com].

Financially, the company is backed by a disclosed $20.8 million Series A round closed in January 2025, with investors including OTB Ventures and eCAPITAL, signaling institutional validation of its venture-scale SaaS model [Crunchbase, Retrieved 2026]. Over the next 12-18 months, the key indicator to watch will be the expansion of its named, multi-party data networks like the ADMET Network, which includes founding members Lundbeck, Orion Pharma, Recursion, and Servier, as these consortia represent the most concrete validation of its platform's value in a production setting [GEN].

Data Accuracy: GREEN -- Core company claims and funding details corroborated by primary sources and multiple press reports.

Taxonomy Snapshot

Axis	Classification
Stage	Series A
Business Model	SaaS
Industry / Vertical	Deeptech
Technology Type	AI / Machine Learning
Geography	Western Europe
Growth Profile	Venture Scale
Founding Team	Co-Founders (2)
Funding	Series A (total disclosed ~$20,800,000)

Company Overview

PUBLIC Apheris was founded in Berlin in 2019 by Robin Röhm and Michael Höh, a pairing that reflects the company's dual focus on life sciences domain expertise and complex technical infrastructure [TechCrunch, Jan 2025]. The founding narrative, as detailed on the company's website, stems directly from Röhm's prior startup experience, where he reportedly lost customers because regulatory constraints prevented the centralization of data necessary for collaborative AI work [apheris.com]. This firsthand encounter with the data bottleneck in regulated industries became the catalyst for Apheris, which was built to enable model training across distributed, sensitive datasets without moving them.

The company's early development culminated in a public platform launch in September 2021, which it framed as a mission to unlock innovation by enabling collaboration on distributed data [apheris.com]. A significant operational milestone followed in January 2025 with the close of a $20.8 million Series A financing round, a capital event that provided a public marker of investor validation and likely funded an expansion of commercial and technical efforts [Crunchbase, Retrieved 2026] [eCAPITAL, Retrieved 2026].

Data Accuracy: GREEN -- Founding date and team confirmed by TechCrunch and company website; funding round corroborated by multiple independent sources.

Product and Technology

MIXED

Apheris sells a federated computing infrastructure, not a single model. The core product, the Apheris Gateway, is an end-to-end platform designed to let life sciences organizations train machine learning models on distributed, sensitive datasets without moving the data to a central location [apheris.com]. The company's framing is that of enabling "life science data networks," where participants can collaborate across organizational boundaries while maintaining strict governance, security, and privacy controls over their proprietary or regulated data [apheris.com].

The platform's functionality is described for several key use cases. It is designed for secure local use, allowing a team to benchmark models on their own in-house datasets. It also enables controlled customization of shared models and provides access to models that have been trained across aggregated industry data [apheris.com]. Specific technical components include a Command Line Interface (CLI) for managing datasets, compute specifications, and jobs, as well as an Apheris Statistics package for privacy-sensitive federated statistical analysis [apheris.com]. The technology stack is built to integrate with existing cloud infrastructure; the company is an AWS Technology Partner and has demonstrated federated learning workflows on NVIDIA DGX Cloud [Amazon Web Services] [apheris.com].

Public deployments illustrate the product's application. The most detailed is the ADMET Network, a federated initiative for collaborative ADMET (Absorption, Distribution, Metabolism, Excretion, Toxicity) model training where founding pharmaceutical members can fine-tune a base model within their own secure environments [GEN Edge]. Apheris also powers a separate AI Structural Biology (AISB) Network and has announced a collaboration to fine-tune the OpenFold3 protein prediction model using proprietary pharma data [apheris.com]. These networks represent the product's realized form: a governed, multi-party computational environment for drug discovery AI.

Data Accuracy: GREEN -- Product claims and technical details are confirmed by the company's own website and blog, with network deployments corroborated by third-party press.

Market Research

PUBLIC The market for federated computing in drug discovery is defined by a fundamental tension: the immense value of proprietary data for training AI models and the legal and competitive barriers to sharing it.

Third-party market sizing for federated learning infrastructure is nascent, but the addressable market can be triangulated from adjacent sectors. The broader AI in drug discovery and development market was valued at $1.2 billion in 2023 and is projected to grow at a compound annual growth rate of 29.1% through 2030, according to a report by Grand View Research [Grand View Research, 2024]. Separately, the global federated learning solutions market was estimated at $127 million in 2022 and is forecast to reach $210 million by 2028, according to research cited by MarketsandMarkets [MarketsandMarkets, 2023]. While these figures represent broader categories, they provide an analogous scale for the specialized niche Apheris targets, which sits at their intersection.

Demand is propelled by several converging tailwinds. The primary driver is the escalating cost and complexity of drug R&D, which creates pressure to use AI for efficiency gains. A secondary, equally powerful driver is the tightening global regulatory environment for data privacy and sovereignty, exemplified by GDPR in Europe and similar frameworks worldwide, which makes traditional data centralization for model training increasingly impractical [GEN, 2024]. The company's own materials cite the formation of industry-specific data networks, like the ADMET Network it hosts, as evidence of a shift towards collaborative but controlled data utilization [apheris.com].

Key adjacent markets include traditional centralized AI/ML platforms (e.g., cloud-based MLOps) and data clean room solutions, which offer alternative approaches to secure data collaboration but lack the federated learning paradigm's core promise of keeping raw data entirely local. The regulatory macro force is a defining characteristic of this sector, not merely a backdrop; it functions as a structural barrier protecting established solutions that can navigate it.

AI in Drug Discovery (2023) | 1200 | $M
Federated Learning Solutions (2022) | 127 | $M

The available sizing data, while not specific to life sciences federated infrastructure, illustrates the substantial growth trajectories of the two converging markets Apheris operates within. The nearly 30% CAGR projected for AI in drug discovery underscores the sector's appetite for technological solutions.

Data Accuracy: YELLOW -- Market sizing figures are drawn from third-party analyst reports (Grand View Research, MarketsandMarkets) and provide an analogous, not direct, view of the target niche. Demand drivers are corroborated by industry coverage and company statements.

Competitive Landscape

MIXED Apheris competes in a nascent but increasingly crowded market for federated learning infrastructure, where its primary differentiation is a vertical focus on the stringent data governance and collaborative workflows of pharmaceutical R&D.

Company	Positioning	Stage / Funding	Notable Differentiator	Source
Apheris	Federated infrastructure for AI training on sensitive pharma data.	Series A (~$20.8M total).	Vertical focus on life sciences; operates federated data networks (e.g., ADMET Network).	[apheris.com], [TechCrunch, Jan 2025]
Rhino Health	Federated learning platform for healthcare data, including imaging.	Venture stage.	Strong focus on medical imaging and clinical trial data; originated from Harvard Medical School.	[Competitor list]
Owkin (Substra platform)	Federated learning for biomedical research and drug discovery.	Later stage (Series B+).	Owns the Substra open-source framework; has established pharma partnerships (e.g., Bristol Myers Squibb).	[Competitor list]
NVIDIA (FLARE platform)	Open-source federated learning framework (NVIDIA FLARE).	Public company.	Deep integration with NVIDIA's AI stack and hardware; horizontal, community-driven tooling.	[Competitor list]
Sherpa.ai	Enterprise federated learning and privacy-preserving AI platform.	Venture stage.	Broader horizontal enterprise focus across finance, telecom, and healthcare.	[Competitor list]

The competitive map segments into three tiers. The first comprises horizontal infrastructure providers like NVIDIA, whose FLARE framework offers a foundational, open-source toolkit but requires significant integration work for regulated industries. The second includes cross-sector enterprise platforms such as Sherpa.ai, which offer broader privacy-preserving AI solutions but lack the domain-specific data connectors and compliance assurances required for drug discovery. The third and most direct tier consists of life-sciences-native federated learning specialists. Here, Rhino Health and Owkin represent Apheris's most immediate challengers, each with established academic or commercial footholds in clinical and biomedical data collaboration.

Apheris's current edge appears to be its execution on the network model. While competitors offer platforms, Apheris is actively building and operating multi-party federated data networks, such as the ADMET Network with named pharmaceutical members [GEN]. This moves the company beyond selling tooling to orchestrating industry consortia, creating a potential lock-in through ecosystem effects. The company's partnership with Amazon Web Services as a technology partner for Amazon Bio Discovery also provides a credible, though not exclusive, channel [Amazon Web Services], [bio-itworld.com, 2026]. This edge is durable if network participation drives recurring platform usage and proprietary protocol development, but perishable if larger competitors replicate the network model or if participants churn due to platform limitations.

The exposure for Apheris is twofold. First, it faces competition from well-capitalized horizontal players. NVIDIA can use its dominant hardware and software ecosystem to make FLARE an industry standard, potentially commoditizing the infrastructure layer. Second, within the vertical, Owkin possesses a significant first-mover advantage with its Substra platform and a deeper bench of published research and pharmaceutical alliances. Apheris does not currently own a critical data moat or proprietary base model; its value is orchestration. If a competitor's platform proves more interoperable or cost-effective for the same collaborative workflows, Apheris could be disintermediated.

The most plausible 18-month scenario hinges on adoption momentum within specific therapeutic areas or consortium models. A 'winner' scenario for Apheris would see its ADMET Network and the AI Structural Biology (AISB) Network becoming the de facto standards for federated pre-clinical research, attracting a critical mass of participants that makes switching costly. A 'loser' scenario would materialize if a major pharmaceutical partner standardizes on a competing platform like Owkin's Substra for internal federated projects, relegating Apheris to a niche player and stunting its network growth. The verdict in the Analyst Notes will turn on which of these paths gains more traction.

Data Accuracy: YELLOW -- Competitor positioning inferred from category descriptions; specific funding and stage details for competitors are not fully corroborated by independent public sources.

Opportunity

PUBLIC Apheris is positioned to capture the economic upside of a pharmaceutical industry forced to collaborate on AI without surrendering its most valuable asset, proprietary data.

The headline opportunity is to become the default federated computing infrastructure for the global life sciences industry. The company's core proposition, enabling collaborative AI training across organizational boundaries while keeping data physically local, directly addresses a structural and regulatory bottleneck in drug discovery [apheris.com]. The evidence that this outcome is reachable, not merely aspirational, lies in the formation of multi-party industry consortia on its platform. The launch of the ADMET Network, with founding members including Lundbeck, Orion Pharma, Recursion, and Servier, demonstrates that competing pharmaceutical firms are willing to adopt a shared, neutral platform for a critical, pre-competitive research function [GEN Edge]. This early adoption by regulated enterprises validates the technical and governance model, providing a foundation from which Apheris can standardize workflows for other high-value use cases like structural biology and clinical trial analytics.

Growth Scenarios The path from a validated platform to a category-defining business hinges on a few concrete expansion scenarios.

Scenario	What happens	Catalyst	Why it's plausible
Platform Standardization	Apheris Gateway becomes the mandated technical layer for all new industry data consortia.	A major regulatory body or industry group (e.g., IMI, TransCelerate) endorses or adopts the platform for a flagship project.	The company is already an AWS technology partner and is cited as providing AI models for the Amazon Bio Discovery initiative, indicating recognition by a major cloud hyperscaler [Amazon Web Services, aboutamazon.com].
Vertical Expansion into Healthcare	The infrastructure is adapted for hospital systems to collaborate on diagnostic AI models using sensitive patient data (e.g., medical imaging).	A partnership with a large hospital network or medical device company to launch a federated research network.	The platform's cited ability to handle medical imaging data types like DICOM shows product readiness for adjacent healthcare verticals [apheris.com].
Embedded Infrastructure	Apheris's federated orchestration layer is white-labeled or embedded within the cloud service offerings of AWS, Google Cloud, or NVIDIA.	A formal co-sell or go-to-market partnership with a cloud provider's life sciences division.	The technical integration work is evidenced by published collaborations on federated learning with NVIDIA DGX Cloud [apheris.com].

What compounding looks like is a classic network effect, but applied to data ecosystems rather than individual users. Each new consortium or large enterprise that adopts the Apheris Gateway increases the platform's value for the next participant by establishing proven governance templates, security protocols, and legal frameworks for data collaboration. Success in one domain, like ADMET prediction, creates a reference architecture that lowers the activation energy for launching a network in another, such as federated protein folding. The company's blog post on fine-tuning the OpenFold3 model with proprietary pharma data illustrates this dynamic: an initial technical proof-of-concept with academic partners can evolve into a broader industry collaboration, attracting more proprietary datasets and improving model performance for all participants [apheris.com]. This creates a data moat not by centralizing data, but by becoming the indispensable and trusted conduit through which decentralized data generates value.

The size of the win can be framed by looking at the valuation of public companies that own critical, hard-to-replicate infrastructure in adjacent data-intensive sectors. For example, Snowflake, which revolutionized centralized data sharing, reached a market capitalization exceeding $50 billion at its peak. A successful execution of the Platform Standardization scenario, where Apheris becomes the essential federated data layer for life sciences, could support a valuation in the low-to-mid single-digit billions, representing a significant multiple on its current funding. This outcome is contingent on capturing a material portion of the estimated multi-billion dollar market for AI in drug discovery and development, a figure frequently cited by industry analysts. It is a scenario, not a forecast, but it defines the scale of the opportunity the company is attempting to address.

Data Accuracy: YELLOW -- Core product claims and the ADMET Network launch are confirmed by company and press sources. The Amazon partnership is cited by Amazon but specific commercial terms are not public. Growth scenarios are extrapolations based on cited technical capabilities and early market signals.

Sources

PUBLIC

1. [TechCrunch, Jan 2025] Apheris rethinks the AI data bottleneck in life science with federated computing | https://techcrunch.com/2025/01/02/apheris-rethings-the-ai-data-bottleneck-in-life-science-with-federated-computing/

2. [apheris.com] Superior drug discovery models | https://www.apheris.com/

3. [apheris.com] About Apheris | https://www.apheris.com/company

4. [apheris.com] Build & join life science data networks | https://www.apheris.com/product

5. [apheris.com] Advancing protein prediction with Federated Learning on NVIDIA DGX Cloud | https://www.apheris.com/resources/blog/advancing-protein-prediction-with-federated-learning-on-nvidia-dgx-cloud

6. [apheris.com] AlQuraishi lab’s OpenFold3 to Be Fine-Tuned with Pharma Industry Data in a Secure AI Collaboration Powered by Apheris | https://www.apheris.com/resources/blog/alquraishi-lab-s-openfold3-to-be-fine-tuned-with-pharma-industry-data-in-a-secure

7. [Crunchbase, Retrieved 2026] Apheris Company Profile | https://www.crunchbase.com/organization/apheris

8. [eCAPITAL, Retrieved 2026] eCAPITAL invests in Apheris | https://www.ecapital.vc/en/news/ecapital-invests-in-apheris/

9. [GEN Edge] ADMET Predictions Get AI Boost, Federated Data Network Unites Pharma | https://www.genengnews.com/gen-edge/admet-predictions-get-ai-boost-federated-data-network-unites-pharma/

10. [Amazon Web Services] AWS Partner: Apheris | https://aws.amazon.com/partners/find/partnerdetails/?id=0010h00002FJq4FAAT

11. [bio-itworld.com, 2026] Amazon Bio Discovery Partners | https://www.bio-itworld.com/news/2026/01/15/amazon-bio-discovery-partners

12. [Grand View Research, 2024] AI in Drug Discovery Market Size Report, 2024-2030 | https://www.grandviewresearch.com/industry-analysis/artificial-intelligence-in-drug-discovery-market

13. [MarketsandMarkets, 2023] Federated Learning Solutions Market | https://www.marketsandmarkets.com/Market-Reports/federated-learning-market-166699074.html