This grant aims to accelerate research and development of AI foundation models tailored for scientific needs. The primary focus is on four specific domains: materials science, climate change science, environmental pollution science (including PFAS), and agricultural sciences. The goal is to advance AI technology for scientific discovery, contribute to foundation models, and support open-source science.
Explicit identification of target recipient type and size: Primarily research organizations, universities, and potentially companies involved in R&D, typically as part of multi-beneficiary consortia. There are no explicit size restrictions, but collaborative projects are expected.
MUST state if grant is 'SECTOR-SPECIFIC' or 'SECTOR-AGNOSTIC': SECTOR-SPECIFIC
Geographic scope and any location requirements: Applicants must be established in an EU Member State or a country associated with Horizon Europe. Non-EU/non-Associated Countries may also be eligible if specific funding provisions are in place.
Key filtering criteria for initial grant screening: Focus on AI foundation models, specific scientific domains (materials, climate, environmental pollution, agriculture), and a strong research and innovation component (RIA).
Grant frequency and program context: This is a specific topic under Horizon Europe, Cluster 4, Industry, Digital Technologies, for the year 2025. It is part of the broader GenAI4EU initiative and the EU's Apply AI strategy.
Financial Structure
Total budget for this specific topic (HORIZON-CL4-INDUSTRY-2025-01-DIGITAL-61) is 30,000,000 EUR.
The expected number of grants to be awarded for this topic is 5.
The grant amount per successful project (individual contribution) is fixed at 6,000,000 EUR.
The funding will be provided in the form of lump sum contributions, meaning payments are based on the completion of predefined work packages, not on actual incurred costs.
Eligible costs include categories such as personnel, subcontracting, purchase costs (travel, equipment, other goods/services), and other specific costs (e.g., financial support to third parties, internally invoiced goods/services, transnational access to research infrastructure).
Only costs that would normally be considered eligible under standard Horizon Europe actual cost grants can be included in the lump sum calculation.
Indirect costs are calculated using a flat rate of 25% of the direct costs (excluding certain categories) and are integrated into the lump sum.
A co-financing principle applies, meaning the total estimated costs of the action must be greater than the estimated Union contribution.
Payments are linked to the proper implementation of work packages. If conditions for a work package are not met, that portion of the lump sum is not paid, but may be paid in a subsequent period if conditions are then met.
Between 5% and 8% of the total lump sum is retained as a contribution to the Mutual Insurance Mechanism.
Eligibility Requirements
Organization Type & Legal Status
Eligible organizations include public and private entities established in EU Member States or Horizon Europe Associated Countries.
Organizations typically include universities, research organizations, and enterprises (including SMEs).
Projects are expected to be collaborative, requiring the formation of a consortium involving multidisciplinary teams.
Geographic Requirements
Applicants must be based in one of the EU Member States or a country associated with the Horizon Europe Programme.
Specific provisions may allow for funding to participants from certain non-EU/non-Associated Countries, as detailed in Annex B of the Work Programme General Annexes.
Technical & Operational Capacity
Applicants must demonstrate access to high-quality, multimodal data necessary for developing foundation models in their chosen scientific domain.
Capacity to secure and utilize significant computational resources for model training, evaluation, and inference is required.
Expertise in developing robust and reliable AI models, including the ability to integrate domain-specific knowledge.
A plan to make the developed models open, including source code and training datasets, must be in place.
Consortium Requirements
Formation of a consortium is mandatory for this Research and Innovation Action (RIA).
Consortia should ideally include both AI scientists and domain-specific scientists (materials, climate, environmental, agricultural) and, where relevant, experts in Social Sciences and Humanities (SSH) to address legal and ethical aspects.
Application Process
Application Process
Submission Deadline: The final deadline for submitting proposals is 2025-09-23 00:00:00+00.
Submission Window: The submission session opened on 2025-05-22.
Submission Method: Applications must be submitted through the Electronic Submission Service on the Funding & Tenders Portal. This is a single-stage submission process.
Required Documentation: Applicants must use the standard application form specific to Research and Innovation Actions (HE RIA, IA). A detailed budget table (HE LS) for lump sum grants is also required, breaking down estimated costs by work package, beneficiary, and affiliated entity.
Application Content: Proposals must describe in detail the activities covered by each work package and all related resources. A declaration that estimated budgets followed the applicant's own accounting practices is necessary.
Evaluation and Award
Evaluation: Proposals are evaluated by external independent experts. Financial experts will specifically assess the budget estimates based on benchmarks to ensure they are appropriate for the proposed activities.
Timeline: The indicative timeline for evaluation and grant agreement preparation is described in Annex F of the Horizon Europe Work Programme General Annexes.
Support and Resources
Guidance: Comprehensive guidance is available through the Horizon Europe Programme Guide, the Funding & Tenders Portal Online Manual, and a dedicated guide on 'Lump sums - what do I need to know?'.
Helpdesks: Assistance can be obtained from the Research Enquiry Service, National Contact Points (NCPs) in EU and associated countries, the Enterprise Europe Network (EEN), and the IT Helpdesk for technical issues.
Partner Search: The Funding & Tenders Portal offers a partner search tool to help identify potential collaborators for your consortium.
Evaluation Criteria
Overall Assessment Criteria
Proposals will be evaluated based on three main criteria: excellence, expected impact, and quality and efficiency of the implementation.
Specific Award Criteria and Portfolio Balance
To ensure a balanced portfolio of foundation models across various scientific disciplines, grants will be awarded not only based on ranking but also with consideration for the following distribution, provided all applications meet the necessary thresholds:
At least two projects will be selected within the materials science domain (Domain A).
At least one project will be selected within the climate change science domain (Domain B).
At least one project will be selected within the environmental pollution science domain (Domain C).
At least one project will be selected within the agricultural sciences domain (Domain D).
Technical and Scientific Quality
Innovation: Proposals should demonstrate innovative approaches to developing AI foundation models tailored for scientific needs.
Methodology: The proposed research methodology, including data curation, quality control procedures, and model architecture robustness, will be critically assessed.
Adaptability: The ability of the foundation model to be adapted to various subtasks and scientific problems within the chosen domain is a key factor.
Impact and Sustainability
Expected Outcomes: Clear contribution to advancing AI technology for science, bridging knowledge gaps, and supporting open-source and open science practices.
Use Cases: Identification and detailed description of at least four possible use cases and scientific challenges that the model can address.
Risk Mitigation: Identification and assessment of potential risks associated with the misuse of the foundation model.
Sustainability Plan: A clear plan for making the model public, maintaining it, evolving it, and promoting its use within the scientific community on a regular basis.
Compliance & Special Requirements
Regulatory and Ethical Compliance
General Compliance: Applicants must adhere to the general conditions outlined in Annex B of the Horizon Europe Work Programme General Annexes and the EU Financial Regulation 2024/2509.
Data Protection and Privacy: Proposals should involve expertise from Social Sciences and Humanities (SSH) where appropriate, to address issues related to data privacy, sharing agreements, and compliance with relevant regulations.
Ethical Standards: Projects are subject to checks on ethics and research integrity throughout their implementation.
Technical and Operational Requirements
Open Science and Data Sharing: A core requirement is to make the developed foundation models openly available, including source code, and where possible, training datasets and other associated assets, to ensure full reusability by the scientific community.
Data Standards: Projects are expected to contribute to efforts to establish common standards for data formats, metadata, taxonomies, and ontologies.
Risk Management: Proposals must identify and assess potential risks of misuse associated with the developed foundation models.
Strategic and Collaborative Aspects
Strategic Alignment: This topic is an integral part of the GenAI4EU initiative and the broader Apply AI strategy, aiming to bolster Europe's leadership in AI development and adoption.
International Collaboration: International cooperation is encouraged, particularly with initiatives that offer reciprocal benefits for the EU, such as the Trillion Parameter Consortium (TPC).
Synergies: Collaboration with established infrastructures (e.g., WeatherGenerator project) and other relevant EU-funded projects (e.g., HORIZON-INFRA-2025-01-EOSC-06) is encouraged.
Gender Dimension: While not a mandatory requirement for analysis within the research and innovation content, adherence to broader EU policies on gender equality is generally expected in project implementation.
Grant Details
ai
foundation models
generative ai
artificial intelligence
materials science
climate change
environmental pollution
pfas
agricultural science
open source
open science
research and development
innovation
digital technologies
horizon europe
eu funding
scientific research
lump sum
consortium
multidisciplinary
data science
computational resources
ethical ai
social sciences and humanities
sdg
AI Foundation models in science (GenAI4EU) (RIA)
HORIZON-CL4-INDUSTRY-2025-01-DIGITAL-61
Horizon Europe
UNIVERSITY
NGO
ENTERPRISE
SME
OTHER
AT
BE
BG
HR
CY
CZ
DK
EE
FI
FR
DE
GR
HU
IE
IT
LV
LT
LU
MT
NL
PL
PT
RO
SK
SI
ES
SE