EU Funding

Artificial intelligence projects on EU funds

For the first time since the PNRR calls closed, companies building AI solutions have a major call open — with budgets to match their technical ambitions.

Northdan Soft is a software supplier, not a funding consultancy, and is not affiliated with any managing authority or development agency. Programme information is indicative — always check the conditions in the programme's official applicant guide.

Artificial intelligence projects enjoy in 2026 their clearest funding window in years: POCIDIF 2.1, with applications between 30.06 and 30.09.2026, offers non-repayable aid of 200,000 up to 1,500,000 EUR exclusively to SMEs holding an IT&C CAEN code that develop an innovative software solution. The guide treats AI as a smart-specialisation subdomain and states that particular attention will go to trustworthy artificial intelligence systems, defined as technically robust, safe and transparent systems, able to explain the decisions they take and to ensure non-discrimination, diversity and fairness. Meanwhile, the PNRR call for advanced digital technologies (500,000–3,000,000 EUR, with AI on its list) is closed and in implementation, proof that the state funds this direction consistently. A credible AI project needs more than a fashionable word in its title: it needs a real problem, data, a realistic architecture and a staged development plan. This is where Northdan Soft comes in — we build the technical component of the file: the description of the solution and the models used, the budget by eligible cost category, the draft contract, then the actual development and the reimbursement documents.

Let’s talk about your project

Message us on WhatsApp or send an email — you talk directly to a developer.

Message us on WhatsApp Send an email

office@northdan.com · +40 752 070 247

How we help

AI architecture explained for evaluators

We translate models, data and infrastructure into measurable deliverables and stages, without jargon that hides a lack of substance.

A realistic budget for development and infrastructure

Training and running models costs money; the budget separates development, computing resources and integration so everything stays reimbursable.

Staged development with demonstrable results

Prototype, validation on real data, production version — each stage produces evidence for the progress reports.

POCIDIF 2.1 and the AI funding context of 2026

The call now open addresses IT&C-sector SMEs developing their own products: if the AI solution is meant for sale, not just internal use, you sit inside the programme's target. Large amounts — up to 1,500,000 EUR for a software solution, the 3,000,000 EUR figure in the press belonging to hardware products manufactured at the beneficiary — bring matching expectations at evaluation: team, commercial exploitation plan, indicators.

The PNRR call for advanced technologies already proved the market's appetite: projects with AI, cloud, IoT and blockchain are now in execution. Whoever missed that train has a second chance in POCIDIF 2.1, but with the 30.09.2026 submission deadline, preparation time is short.

What a serious AI project means, from a developer's standpoint

Before promising models, we verify the data: does it exist, is it sufficient, can it be used legally? Then we define the use case — prediction, classification, generation, decision automation — and only then choose the technology. This order shows in the technical offer and makes the difference at evaluation.

During implementation we work iteratively, reporting performance metrics at every stage, and at the end we hand over the solution with complete reimbursement documentation. We do not write the grant application, we do not issue eligibility verdicts and we cannot be your entire technical team: the guide requires at least 20% of the aid for the core activity to go to the beneficiary's own staff.

Frequently asked questions

My company is not in IT but wants to use AI internally. Can that be funded?

Not through POCIDIF 2.1, which requires an IT&C CAEN code and product development. Internal-use AI components can, however, appear in other digitalization projects, per the official applicant guide of each call — usually as software or development services, not as research.

Which AI costs can typically go into the budget?

Software development of the solution, computing resources for training and inference, integration with existing systems, testing and documentation. Data acquisition or model licensing is treated differently from guide to guide, so it gets checked case by case before budgeting.

Is a prototype mandatory before applying to POCIDIF 2.1?

The official applicant guide sets the exact maturity requirements. From our experience as developers, a prototype or at least a documented technical validation strengthens the file considerably, because it turns a promise into a calculated risk — and we can build precisely that prototype together.