Second call for innovation studies for the development of generative AI models

The FFplus project is opening its second call for proposals aimed at innovation studies focused on developing generative AI models, specifically targeting small and medium-sized enterprises (SMEs) and startups proficient in generative AI, professional software development, and data processing. The program encourages these organizations to leverage high-performance computing (HPC) resources, with projects needing to utilize European HPC infrastructure to create and customize AI models.

The total funding available for this call is 4,000,000 euros, with an application deadline set for November 26, 2025. Each innovation study can receive up to 300,000 euros, with a maximum of 200,000 euros allocated to the main participant, an SME or startup. Supporting participants may receive additional funding under specific conditions. The initiative emphasizes the development and validation stage of projects, particularly foundation and large language models, and disallows model exploitation or deployment.

Eligible applicants must be established in EU Member States, EEA countries, or countries associated with the Digital Europe Programme. Each application can consist of a main participant and up to two supporting partners, limited to three total. The application process is a single-stage submission through the Fortissimo project website.

The funding mechanism is structured as a grant covering 100% of eligible direct costs, with no financial contribution required from applicants. Successful proposals will be evaluated based on their impact, excellence, and implementation strategy, with a threshold score of 10 required for acceptance.

Proposals must be submitted electronically and include detailed administrative information as well as specific innovative study descriptions. Adherence to format and submission guidelines is critical, as any deviations may lead to rejection.

The call supports the European objectives around digital sovereignty and competitiveness in the AI sector. By providing full funding and access to advanced HPC resources, it aims to foster innovation among European SMEs and reduce dependency on non-European AI solutions. Proposals must demonstrate a clear business case for developing new generative AI models, highlighting the necessity of their work relative to existing solutions.

Overall, the FFplus initiative is positioned as a key driver for empowering European tech innovation, endorsing trustworthy AI development, and scaling SMEs in the generative AI domain.

The FFplus project is announcing its second call for proposals for innovation studies focused on the development of generative AI models. This initiative is driven by the business needs of SMEs and Start-ups with expertise in generative AI, professional software development, and data processing. The innovation studies are required to utilize large-scale European High-Performance Computing (HPC) resources, including pre-exascale and exascale supercomputers, to develop and customize generative AI models, such as foundation and large language models. This call is complementary to the open call for business experiments addressing the uptake of HPC by SMEs (Identifier FFplus_Call-2-Type-1), and SMEs can only participate in one of the two types of sub-projects.

The opening date for submissions is September 10, 2025, and the deadline for submissions is November 26, 2025, at 17:00 Brussels time. The expected duration of participation is 10 months, with a targeted commencement date of April 1, 2026. The total funding available for this call is 4,000,000.00 €. The project acronym is FFplus, and the full name of the EU-funded project is Fortissimo Plus. The grant agreement number is 101163317, and the topic is DIGITAL-EUROHPC-JU-2023-SME-01-01 - Supporting competitiveness and innovation potential of SMEs.

All submissions must be made electronically via the proposal submission tool on the Fortissimo website. The submission must include a spreadsheet containing administrative information and a PDF document (maximum size 5.0 MB) that adheres to the provided proposal structure instructions. Proposals must be submitted in English and comprise two parts: Part A (administrative information) and Part B (the body of the proposal). Part A is submitted as a spreadsheet with tables for administrative data, including a list of participants with their Participant Identification Code (PIC) issued by the European Commission, valid address, telephone number, and email contact data. While a PIC is not mandatory, registration can be done via the provided link. Only requested information should be included in Part A, as extraneous information will be deleted.

Part B of the proposal must not exceed 12 pages (excluding the cover page and scientific literature references). The text should be in no smaller than 11-point Arial font, including text within images. Proposals exceeding the page limit will be rejected. Proposal templates for both Part A and Part B, including embedded Excel spreadsheets for the budget, are available on the project website and must be followed.

The evaluation criteria are:

Impact: Prospects for innovation, commercial viability, societal relevance, vision of success, value creation/proposition, alignment with call objectives, SME business models, and exploitation plans.
Excellence: Conceptual and technical excellence, including conceptual soundness, cohesiveness, articulation of plans for bridging gaps, clear definition of technical requirements, justification of technology choices, articulation of performance metrics for model evaluation, scaling, and optimization, activities for establishing baseline performance, and ensuring experiment reproducibility.
Implementation: Quality of the project's work plan and data management plan, distribution of resources, capacity of the applicant(s) to carry out the proposed work, and justification for computation resources required.

Each criterion will be assigned a score between 0 and 5. The overall acceptance threshold is 10 (summed over all criteria), with a minimum score of 3 for each criterion. All criteria are equally weighted. In case of a tie, proposals will be ranked based on the individual criteria scoring in the following priority: Impact, Excellence, Implementation, and finally total requested funding.

The evaluation will be conducted by independent, external experts using a consensus review process supported by the FFplus beneficiaries. Each proposal will be assigned two experts who will produce individual assessment reports. A moderator will then work with the evaluators to create a consensus assessment report, which will be turned into an evaluation summary report shared with the proposing consortium.

Adherence to the proposal format and page limits is crucial for evaluation. Failure to comply with funding and eligibility restrictions or proposal format instructions will lead to immediate rejection. Proposers will receive an evaluation summary report with the consolidated findings of the experts and a decision on the selection procedure, which is final, and no appeals process will be provided.

Successful consortia will be included as sub-projects upon concluding a funding agreement with the University of Stuttgart, the FFplus project coordinator. This agreement is non-negotiable and available on the project website. By submitting a proposal, proposers acknowledge this agreement.

FFplus is funded by the EuroHPC Joint Undertaking (EuroHPC JU) under the DIGITAL-EUROHPC-JU-2023-SME-01 action, which aims to empower SMEs with advanced computational capabilities based on HPC. The project will use open calls to select business experiments and innovation studies, funded through Financial Support for Third Parties (FSTP). FFplus will provide over €24 million to sub-projects through six open calls: three for SMEs using HPC to improve their business (business experiments) and three for SMEs/Start-ups in generative AI (innovation studies).

This call specifically targets SMEs and Start-ups proficient in generative AI and HPC, supporting their innovation potential by leveraging new generative AI models, such as Large Language Models (LLMs). For sub-projects funded under FSTP, an agreement will be concluded between the innovation study consortium and the FFplus Coordinator, including provisions for avoiding conflicts of interest. Before funding agreements, SMEs must document their self-assessment of adherence to EU guidelines for trustworthy AI and their SME status according to European Commission definitions.

Third Parties will receive 100% funding of incurred, eligible direct costs necessary for the completion of experiment activities; no indirect costs or overheads will be funded. The FFplus project receives funding based on a Grant Agreement following the regulations of the Digital Europe Programme and the eligibility rules of that Grant Agreement will apply for the direct costs arising in the sub-projects.

The funding agreement will define the payment schedule, with payments made in tranches linked to the delivery of reports and outputs. Sub-project participants will receive an advance payment at the start, followed by subsequent installments. SMEs may participate in more than one tranche of innovation studies if their initial study is successful. Innovation studies from the first tranche will be informed by September 30, 2025, if they are eligible for extension funding. A necessary condition for funding is compliance with the FFplus requirement for reporting results achieved up to the end of the 7th month to enable an evaluation of the impact potential by external experts. Main participants from the top 70% of successful innovation studies will be eligible to submit a proposal for an extension of the study.

The following funding conditions and eligibility criteria apply:

A main participant must be an SME or a Start-up, with supporting participants assisting in completing the innovation study activities. Applications must be submitted by the main participant, who must provide a business case/challenge and optionally up to two supporting participants, each with a clearly defined role. The total number of consortium partners is limited to three. Only organizations established in an EU Member State or listed EEA countries and countries associated with the Digital Europe Programme are eligible for funding. Natural persons are not eligible. FFplus beneficiaries are ineligible to participate. The maximum duration of innovation studies is 10 months, with a maximum of 200,000 € for the main participant SME. The maximum total funding for each innovation study is 300,000 €. Funding limits for supporting participants are a maximum of 150,000 € under this call (FFplus_Call-2-Type-2) and a maximum of 300,000 € over all FFplus innovation studies. SME participation in business experiments and innovation studies is mutually exclusive. At least 50% of the funding should be allocated to the main participant, with any deviations duly justified. For supporting participants, only engineering/technical activities are eligible for funding.

The FFplus project does not provide HPC computing resources itself but expects sub-projects to use HPC resources provided under the EuroHPC JU access schemes, which are free of charge. The use of nationally provisioned HPC resources is considered equivalent. Proposers should refer to the EuroHPC JU website for information about the various access calls, organized in the areas of "supercomputers access" and "AI Factories access". The AI Factories access scheme is expected to be the most appropriate choice. HPC National Competence Centres may aid proposers with the selection of appropriate resources and access schemes. The use of 3rd party commercial HPC resources is permissible when justified, but the costs must be included in the proposal budget. Similarly, the use of resources provided by an organization included within the sub-project consortium as a supporting participant is permissible when suitably justified, based on actual costs, not commercial rates.

The budget modules available for innovation studies include:

Personnel: Costs must be commensurate with the work performed.
Equipment: Depreciation costs only, for specialized equipment necessary for the experiment.
Travel: Must be justified in terms of the necessity for the work plan.
HPC compute capacity: Compute Resources need to be justified. Ideally, EuroHPC systems will be used for the work. A decision not to apply for access to the EuroHPC systems should be duly justified.
Material: Costs for acquiring specialized SW licenses and data sets or collections needed to conduct the experiment.

Sub-contracting costs are not eligible for funding.

Innovation studies will receive support from the project relating to interactions with the project and potential collaborations with other sub-projects, as well as direct support for gaining access to EuroHPC JU-provided computing resources and technical consultation relating to the effective execution of the innovation study work plan.

The structure of Part B of the proposal should be as follows:

Summary (0.5 pages)
Industrial relevance, potential impact (including societal impact) and exploitation plans (3.5 pages)
Description of the work plan, technological/algorithmic approach, and software development strategy (5 pages)
Quality of the consortium as a whole and of the individual proposers (1-2 pages)
Justification of costs and resources (1-2 pages)

A management structure will be established for the successful proposals, but the proposal does not need to contain a description of how the management of the innovation study in the framework of the overall FFplus project will be achieved, but should include tasks for the technical management of the study activities.

The proposed innovation studies are expected to:

Be fully aligned with the FFplus call objectives.
Justify why generative AI is a solution to the business problem, why developing a new model is imperative, and why it could not be addressed sooner, listing obstacles preventing the utilization of existing generative AI models.
Present a vision of success, defining value propositions and the process of value creation.
Define specific objectives achieved through an ML lifecycle: data preparation, model development/engineering, and model evaluation.
Provide a detailed description and demonstrate the availability of a suitable training data set.
Detail the characteristics of the models to be developed, including type, size, hyperparameters, and architecture.
Establish Performance Metrics and a Reproducibility Plan.
Identify potential risks considering EU guidelines for trustworthy AI and present means to address and mitigate them.
Justify Resource Allocation.
Describe the ML lifecycle.
Submit a Comprehensive Data Management Plan.
Support the FFplus project in generating success stories.
Produce a pre-final results and potential impact report by the end of the 7th month.

This FFplus call aims to bolster the competitiveness and innovation potential of SMEs by providing them with the resources to develop and customize generative AI models using European HPC infrastructure. It focuses on supporting innovation studies that address specific business needs and contribute to the advancement of generative AI technologies within the European landscape. The funding is provided to cover direct costs, and the project offers support in accessing HPC resources and technical consultation. The call encourages SMEs and Start-ups to leverage HPC to overcome business challenges and create new opportunities in the field of generative AI.

Eligible Applicant Types: The eligible applicant types are SMEs and Start-ups. A main participant must be an SME or a Start-up. Supporting participants can be other organizations assisting the main participant. Natural persons (individuals) are not eligible. FFplus beneficiaries are ineligible to participate.

Funding Type: The funding type is a grant, specifically Financial Support for Third Parties (FSTP).

Consortium Requirement: A consortium is required. Applications are to be submitted by the main participant (SME or Start-up) who must provide a business case/challenge and optionally (if well justified) up to two supporting participants. The total number of consortium partners is limited to three.

Beneficiary Scope (Geographic Eligibility): Only organizations established in an EU Member State or listed EEA countries and countries associated with the Digital Europe Programme where the association agreement entered into force before the submission deadline are eligible to receive funding.

Target Sector: The target sector is generative artificial intelligence (AI), professional software development, and data processing, with a focus on using high-performance computing (HPC) resources.

Mentioned Countries: EU Member States, EEA countries.

Project Stage: The project stage is development and customization of generative AI models, such as foundation and large language models, using large-scale European HPC resources. The focus is on innovation studies. Model exploitation/deployment/operation is out of scope.

Funding Amount: The maximum funding for the (main participant) SME is 200,000 EUR. The maximum total funding for each innovation study is 300,000 EUR. A supporting participant can receive a maximum of 150,000 EUR under this open call and a maximum of 300,000 EUR over all FFplus innovation studies. The total funding available for this call is 4,000,000 EUR.

Application Type: The application type is an open call.

Nature of Support: Beneficiaries will receive money in the form of a grant, covering 100% of incurred, eligible direct costs necessary for the completion of experiment activities. No indirect costs or overheads will be funded. Direct support for each individual innovation study will also be provided relating to gaining access to EuroHPC JU-provided computing resources and technical consultation relating to the effective execution of the innovation study work plan.

Application Stages: The application process appears to be single-stage, with a single deadline for submissions.

Success Rates: The success rates are not explicitly mentioned, but the top 70% of successful innovation studies from the first tranche may be eligible to submit a proposal for an extension of the study in a subsequent open call.

Co-funding Requirement: Co-funding is not explicitly required, as 100% of eligible direct costs are funded. However, the use of EuroHPC JU resources (provided free of charge) or nationally provisioned HPC resources is strongly encouraged. If 3rd party commercial HPC resources are used, the costs need to be included within the proposal budget.

This FFplus open call, funded by the EuroHPC Joint Undertaking, aims to support SMEs and Start-ups in the field of generative AI by providing financial support and access to high-performance computing resources. The call focuses on innovation studies that develop and customize generative AI models using large-scale European HPC infrastructure. Eligible applicants are SMEs and Start-ups from EU Member States, EEA countries, and countries associated with the Digital Europe Programme. Consortia can consist of up to three partners, including the main applicant (SME or Start-up) and up to two supporting organizations. The maximum funding per innovation study is 300,000 EUR, with a maximum of 200,000 EUR for the main SME participant. The funding covers 100% of eligible direct costs. The application process involves submitting a proposal with administrative information and a detailed description of the innovation study, including its industrial relevance, technical approach, and resource justification. The proposals will be evaluated by independent experts based on impact, excellence, and implementation criteria. The selected consortia will enter into a funding agreement with the University of Stuttgart, the FFplus project coordinator, and receive payments in tranches linked to the delivery of reports and outputs. The call encourages the use of EuroHPC JU resources and provides support for accessing these resources. This initiative seeks to empower European SMEs with advanced computational capabilities, fostering innovation and competitiveness in the generative AI domain.