How Mid-Market Companies Can Fund AI Projects Through Horizon Europe

The Opportunity Nobody's Taking

Horizon Europe is a €95.5 billion R&I program running from 2021 to 2027. It funds everything from breakthrough basic research to market-ready innovation. And while the academic community has mastered it, mid-market companies — companies with 50 to 500 employees, actual customers, and real revenue — are systematically underrepresented in the applicant pool.

This isn't because they're ineligible. It's because the messaging around Horizon Europe skews toward universities and deep-tech startups. The word "research" puts off commercial operators who think they're building products, not papers. That's a misread.

Horizon Europe explicitly targets innovation that moves to market. Several of its programs are designed specifically for companies that already have a product and want to take it to the next level. If your AI project has a defined use case, commercial intent, and measurable impact, you are exactly the kind of applicant these programs were designed for.

Who Actually Qualifies

Horizon Europe is not just for startups or universities. The programs most relevant to mid-market AI companies include SMEs (under 250 employees, under €50M annual turnover) and "mid-caps" (up to 750 employees in certain programs). But even larger companies participate — typically as consortium partners or through specific industrial calls.

For AI specifically, the most important eligibility factors are:

• Innovation level: You must be advancing beyond the state of the art — not just applying existing AI tools. Fine-tuning GPT-4 for your niche doesn't qualify. Building novel architectures, training approaches, or domain-specific models does.
• Commercial pathway: The project needs a credible route to market within the program timeframe. Horizon Europe is not pure research funding — the EIC especially wants to see business models, go-to-market plans, and revenue projections.
• Impact: Your project should contribute to EU strategic priorities — AI sovereignty, industrial automation, climate tech, health, or digital infrastructure. "AI for ad optimization" is a harder sell than "AI for predictive maintenance in manufacturing."
• Organisational capacity: You need the team and infrastructure to manage a grant. Reporting requirements are real. If you've never managed a public grant before, budget for it.

The Programs That Actually Work for AI

There are hundreds of Horizon Europe calls open at any given time. Most are irrelevant to a mid-market AI company. Here are the five worth knowing:

1. EIC Accelerator

This is the flagship program for SMEs commercializing breakthrough innovations. It offers a mixed financing instrument: up to €2.5M in grant funding plus up to €15M in equity investment from the EIC Fund. You can apply for grant-only or grant plus equity.

The key: the EIC is explicitly looking for "deep tech" — but AI qualifies when it's solving a genuine technical challenge, not just a business process problem. Applications are evaluated on innovation (how novel is it?), impact (how big is the market?), and team (can you execute?).

The application is phased. You submit a short application first (proposal + pitch deck + 10-minute video). If you pass, you're invited to a full proposal and then an interview with EIC evaluators. Success rates at the final stage are around 5-8%, but the interview stage acceptance rate is closer to 35%.

2. EIC Pathfinder

For companies working on foundational AI — new model architectures, novel training paradigms, theoretical breakthroughs that could become the next generation of AI infrastructure. Grants of up to €4M for small consortia, up to €10M for larger ones. This is closer to research, but open to companies, not just universities.

If you're building something genuinely novel at the architecture level — not fine-tuning, not RAG pipelines — this is the program to look at.

3. Cluster 4: Digital, Industry and Space

This is where most mid-market AI projects land. Cluster 4 runs dozens of calls per year across themes like AI, data, advanced manufacturing, Industry 4.0, robotics, and digital infrastructure. Grant sizes range from €1M to €10M+ depending on the call, and most require a consortium of 3-8 partners across different EU countries.

The consortium requirement is the part that trips up mid-market companies. You need to build a team of partners — typically including a university or research institute, at least one SME, and ideally one large industrial company. The EU wants to see technology transfer, so having an academic partner that validates your approach is almost always required.

4. Digital Europe Programme (DIGITAL)

Technically separate from Horizon Europe, but often confused with it. DIGITAL funds deployment and scaling of existing digital technologies rather than R&D. If your AI product is proven and you want funding to deploy it across multiple EU member states, roll it out to public sector customers, or build interoperability with other systems — this is the right program. Budget: €7.6 billion for 2021-2027.

5. Marie Skłodowska-Curie Actions (MSCA)

Often overlooked by companies, MSCA funds researcher mobility and training. If you're hiring a PhD-level AI researcher from a university, or sending your technical staff to partner institutions for training, MSCA co-funds the costs. Useful for building the academic credibility you need for other Horizon Europe applications — and for attracting top-tier technical talent with EU co-funding.

The Application Process (What Nobody Warns You About)

Horizon Europe applications are submitted through the EU Funding & Tenders Portal. First step: register your organisation as a Legal Entity Appointed Representative (LEAR). This takes 2-4 weeks and needs to happen before you submit anything.

For most calls, the process is:

1. Find the right call. The Funding & Tenders Portal lists all open calls. Filter by programme, sector, and deadline. AI-relevant calls appear under "Cluster 4: Digital, Industry and Space" and under "European Innovation Council."
2. Read the Work Programme. The Work Programme document for each cluster specifies exactly what is and isn't funded in that call. It's long, dense, and non-negotiable. If your project doesn't match the described scope, reviewers will notice.
3. Build your consortium (if required). For most Cluster 4 calls, you need partners. Use the EU's Partner Search tool on the Funding & Tenders Portal, or reach out directly to relevant universities and industrial associations. Budget 3-6 months for this step alone.
4. Write the proposal. The standard format is a 30-50 page document covering: scientific/technical excellence, impact, and implementation. You'll also complete a financial breakdown, CVs for key personnel, and administrative forms. Budget 4-8 weeks of dedicated writing time, or hire a professional proposal writer.
5. Submit and wait. Evaluation takes 5-8 months for most calls. You'll receive reviewer comments regardless of outcome. If rejected, these are invaluable for your next attempt.

This is the part most companies don't account for: Horizon Europe is a multi-year commitment. The companies that extract the most value treat it as a capability — a standing process that runs alongside the business — rather than a one-off grant application.

Building a Consortium That Works

A bad consortium is worse than no consortium. Partners who don't deliver kill your project and your reputation with reviewers. The best Horizon Europe consortia have three characteristics:

Complementary but non-overlapping roles. Each partner should bring something the others don't have — domain expertise, infrastructure, market access, academic credibility, regulatory knowledge. If two partners are doing the same thing, one of them is redundant and reviewers will ask why.

Clear financial arrangements upfront. Horizon Europe reimburses costs after they're incurred (typically 70-100% of eligible costs depending on the type of organisation). Partners need working capital to fund their share before reimbursement arrives. Many consortium failures are cash flow failures, not technical failures.

A coordinator who owns the process. Someone needs to own project management, reporting, and financial coordination. In most consortia, this is the company that had the original idea and brought the partners together. Budget 0.5-1 FTE for this role throughout the project lifetime (typically 3-4 years).

Aligning With EU AI Priorities

Horizon Europe evaluators are not neutral. They're looking for projects that advance the EU's stated strategic goals. For AI specifically, the EU has identified several priority areas in its AI strategy and the Coordinated Plan on AI:

• Trustworthy AI: Explainability, fairness, robustness, and compliance with the EU AI Act. If your AI project includes work on any of these dimensions, make it central to your application.
• Industrial AI: AI applied to manufacturing, supply chain, energy systems, and infrastructure. The EU is very focused on industrial competitiveness — applications that strengthen European manufacturing or reduce dependency on non-EU AI providers score well.
• AI for sustainability: Climate, energy efficiency, agriculture, and circular economy. The EU AI Act and Green Deal are closely linked in funding priorities. If your AI application has measurable environmental impact, it's a significant differentiator.
• AI sovereignty: This is increasingly explicit in calls. Projects that build European AI capabilities — models trained on European data, hosted in European infrastructure, governed by European companies — are politically preferred right now.

This doesn't mean you need to retool your project to fit these priorities. It means you need to honestly assess which of these your project already serves — and make that connection explicit in your application. Reviewers aren't mind-readers.

What the Numbers Actually Look Like

Horizon Europe reimburses a percentage of your "eligible costs" — the actual costs incurred by your team on the project. What counts as eligible depends on the call, but generally includes:

• Personnel costs (salaries, social security, employer taxes)
• Subcontracting costs (within limits)
• Equipment and infrastructure costs (depreciation, not purchase)
• Travel and accommodation
• Indirect costs (overhead — at a flat rate, usually 25%)

For SMEs in Research and Innovation Actions (RIA), the reimbursement rate is 100%. For Innovation Actions (IA), it's 70% (or 100% for non-profit entities). This means you need to fund 30% of eligible costs from your own budget for IA projects.

The typical budget for a mid-size Cluster 4 project is €4-8 million across the consortium. Your share as the coordinator might be €1-2 million over 3-4 years, reimbursed at 70%, meaning you get back €700K-1.4M against a commitment of €300K-600K of your own resources. The grant is substantial — but so is the commitment.

Common Mistakes (and How to Avoid Them)

Applying to the wrong program. The most common mistake. If your project is already at TRL 7-8 (technology fully functional, approaching commercial deployment), the EIC Accelerator is the right call, not a Cluster 4 research grant. Match your technology readiness level to the program's target range.

Underselling the innovation. "We're applying AI to optimize our business process" doesn't win. "We're developing a novel graph neural network architecture for real-time inference in constrained edge environments, with direct applications to smart grid management" does. The technical ambition needs to be explicit, specific, and defensible.

Ignoring the commercial sections. Horizon Europe is increasingly market-focused. Empty commercial sections — "we will develop a go-to-market plan during the project" — are red flags. You need a defined target market, a realistic business model, and evidence that someone will pay for what you're building.

Treating it as a one-time event. First applications have a ~15-20% success rate for competitive calls. Companies that succeed long-term treat Horizon Europe as an ongoing function, not a one-off project. Allocate internal capacity accordingly.

Not using professional support. The EU has a network of National Contact Points (NCPs) — free advisors in every member state who can help you find the right call, review your draft, and connect you with potential consortium partners. In Portugal, the NCP for digital and industrial innovation is FCT (Fundação para a Ciência e a Tecnologia). Use them. They exist for this.

Getting Started This Quarter

If you're a mid-market company with an AI project worth funding, here's the 90-day starting point:

1. Register on the EU Funding & Tenders Portal and get your organisation's PIC (Participant Identification Code). This takes days, not weeks, and blocks everything else.
2. Browse open calls in Cluster 4 and the EIC. Set alerts for new calls in your domain. The portal lets you filter by keyword, sector, and deadline.
3. Contact your National Contact Point. Book a 30-minute call. They'll tell you whether you're targeting the right programs and what your application would need to succeed.
4. Identify potential consortium partners if your target program requires them. Academic institutions with AI labs are a good starting point — look for ones that publish in your application domain.
5. Write a 2-page concept note before committing to a full application. Share it with your NCP and any potential partners. If you can't write 2 clear pages about the problem, the innovation, and the impact, you're not ready to write 50.

The window that matters is now. Horizon Europe runs through 2027. The major AI calls are concentrated in 2025 and 2026, with a final cluster of high-value calls in 2027 before the next framework programme begins. Companies that build this capability in the next 18 months will be positioned for the next program too.

The money is there. The programs exist for exactly the work you're doing. The barrier is not eligibility — it's willingness to learn a process that most companies haven't bothered to understand.