There is no single figure for how much public money a Luxembourg project can carry, because there is no single scheme. Each instrument publishes its own share of eligible costs and its own ceiling, and they do not agree with each other: R&D project aid reaches 60% of eligible costs for industrial research by a small enterprise but 20% for experimental development by a large enterprise, while SME Packages Digital sits at 70%.
Every one of those figures is a percentage of costs, which is the point of the design: each scheme funds a share and leaves a remainder, and the remainder is where private money goes.
So the useful question is not "what is the stacking limit" but a narrower one: what does each instrument you qualify for actually pay, against which costs. The published rates are below, scheme by scheme; what the tables cannot tell you comes after them.
What national R&D aid pays, by company size
R&D project aid is where most technical budgets start, and its rate is driven less by the project than by which size column the company falls in. The guichet.public.lu R&D project aid page publishes the base rates as a grid of project category against company size:
| Project category | Small enterprise | Medium enterprise | Large enterprise |
|---|---|---|---|
| Industrial research, base rate | 60% | 50% | 40% |
| Experimental development, base rate | 40% | 30% | 20% |
Which column applies is decided by the definition the scheme itself uses — worth confirming before a budget rests on a rate. Company size also appears in guichet's published grid of increases, worth 20 percentage points for a small enterprise and 10 for a medium one; the table above reads straight off the size column, so check on the scheme page which increases apply on top of it.
The scheme has a floor as well as a ceiling: a minimum aid amount of €1,000 for an SME and €100,000 for a large enterprise, which quietly rules out the very small project, where the file costs more to assemble than the aid is worth. Our RDI co-funding guide covers how the categories are drawn.
The early-stage grants, side by side
These are the instruments an early-stage company will actually be choosing between. Each is a separate application with its own eligibility test, and none of them entitles you to the next — treating them as a ladder is the most common sequencing error in this whole area.
| Instrument | Who it is written for | What it pays | Shape |
|---|---|---|---|
| Aide à la primo-création | newly created micro-enterprises in the commerce and craft sectors | up to €12,000, paid as €2,000 a month for up to 6 months | monthly grant |
| Fit 4 Start | company up to 5 years old, team of at least 2, in digital technologies, health or space, or the field specified in the Fit 4 Start call | up to €150,000, in instalments of €50,000, €80,000 and €20,000 | grant plus 6 months of coaching |
| SME Packages Digital | eligible costs of €3,000–25,000 | 70% of eligible costs | cost-share grant |
| Fit 4 Digital | the assessment stage | €5,000 | fixed grant |
| Young Innovative Enterprise | company under 5 years, turnover from €40,000 | up to €1,000,000, co-funding up to 70% | co-funded grant |
Young Innovative Enterprise repays close reading, because its tests are the strictest: a funding plan running up to 3 years, and innovation-certificate criteria that include an R&D spending share of at least 15%. Primo-création trips people for a different reason — it looks at what else you own, capping a holding in another company at 25%.
The debt layer, and why it is capped as a share
Grants are only part of the architecture. SNCI's loan products carry both a euro range and a maximum SNCI share, which by construction leaves a remainder for the company's own funds and for other lenders. That is the structural evidence that these instruments were designed to sit beside private money rather than replace it.
| SNCI product | Range | Maximum SNCI share | Condition on the file |
|---|---|---|---|
| Prêt Création-Transmission | €5,000–250,000 | 40% | minimum equity share of 15% |
| Prêt Entreprises Novatrices | up to €1,500,000 | 35% | company up to 8 years old |
| Prêt Recherche, Développement, Innovation | up to €250,000 | 40% | — |
| Prêt moyen/long terme | €25,000–25,000,000 | — | term up to 15 years |
The equity condition on the creation loan is the sharpest illustration in the system: private money is not something the public instrument tolerates, it is something the product requires before it will lend. SNCI's equipment credit runs from €12,500 to €2,500,000; the full product set is on snci.lu and in our SNCI loans guide.
Where EU money sits, and why it is a separate question
European instruments are a separate system, with their own eligibility, calendar and application route: qualifying for a national scheme tells you nothing about qualifying for a European one. The EIC Accelerator is built as blended finance — a grant component plus direct equity or quasi-equity such as convertible loans — with a grant component up to €2,500,000 running up to 24 months and an investment component between €1,000,000 and €10,000,000.
Read those as separate components, not a single total. The EIC also limits repetition: under the EIC Work Programme 2026, grant-only support can be provided only once to a beneficiary during Horizon Europe, for a maximum of EUR 2.5 million covering activities of TRL 6 to 8. For 2026 the budget is €414,000,000 for EIC Accelerator Open and €220,000,000 for the Challenges.
Horizon Europe works on funding rates rather than ceilings. The general annexes to the 2026–2027 work programme put the maximum at 100% for a research and innovation action and 70% for an innovation action, adding that other funding rates may be set out in the specific call or topic conditions. The rate that binds you is the one in your call, not the one in the annex.
A published rate is not available money
Schemes run on budgets, and budgets run out inside the year, so the sequencing error that costs founders most has nothing to do with eligibility. The EUIPO SME Fund is the clearest illustration: its 2026 budget is €18,000,000, and as of 7 August 2026 the IP Scan and the trade mark and design vouchers are closed to new applications because their funds are exhausted, while the patent and plant variety vouchers remain open.
A founder who read the rate in spring and planned to apply in autumn was planning against a number that had stopped being available. A funding plan therefore needs a checked date beside every line, not just a rate.
The sequences, and where they break
The patterns that recur are phase arguments rather than entitlements, which is exactly where founders misread them.
The early-stage chain puts an acceleration grant first and a young-company scheme after it, because the eligibility tests look at different things — company age, turnover, R&D share. Nothing carries over. You re-qualify from scratch each time, on that scheme's own criteria.
The R&D layer cake separates the development work, the growth phase and the income that eventually arrives, with the IP Box at the far end operating on income rather than on cost. The instruments differ because the cost base differs, not because the company graduated.
The asset-heavy stack places a capped SNCI product, the company's own funds and a bank alongside each other on the same equipment project. A loan capped at a share of the cost is a smaller ask of the bank than the whole of it, and that is the leverage founders under-use.
What the tables cannot tell you
How instruments behave when they meet on the same cost is genuinely scheme-specific, and no honest figure can be given for it here. What each scheme does publish is its own maximum rate, its own cap and its own eligibility test — which is what the tables above carry.
Whether a cost already supported from another source can appear again in a further application, and what that does to the rate of the scheme reading it, is a question to put in writing to the scheme and to an adviser before the budget is built. The same applies in reverse: nothing here says that raising privately changes what a scheme will pay, and nothing here says it does not.
What investors read in an aid file
Better than founders expect, with a condition attached. Non-dilutive money extends runway without touching the cap table, which makes it a term-sheet asset rather than a distraction. The condition is administrative. Every rate in this article is expressed as a share of eligible costs, which tells you what the file will eventually be about: costs, evidenced, mapped to the categories the scheme recognises.
An aid file that cannot produce that mapping on request reads in due diligence as a signal about how the company keeps its records — a far more expensive problem than the grant.
Which is why project boundaries deserve as much attention as rates. The boundary decides which costs are eligible, which scheme they belong to, and whether the same work is described consistently to a ministry and to a lead investor. Most of the errors in our list of funding application mistakes are boundary errors wearing a different hat.
Common questions
Can a project draw on several public schemes at once? No general rate is published here for that, and a page that hands you one without pointing at a scheme condition is guessing. What is published is per-scheme: the maximum rate, the cap and the eligibility test in the tables above. Put the interaction question to each scheme and to an adviser before assuming it.
Does private investment reduce what a scheme will pay? Do not assume it in either direction. The scheme's own conditions govern, and they differ by instrument.
Which instruments are open right now? Check on the day. Rates are published for a period; availability is not the same thing, as the EUIPO vouchers above show.
What is the smallest project worth applying for? R&D project aid sets a minimum aid amount of €1,000 for an SME, and SME Packages Digital operates on a published band of eligible costs. Below the floor, the effort sits on the wrong side of the arithmetic.
Where does the IP Box fit? On income rather than on cost, under article 50ter L.I.R. — and not flatly: the 80% exemption applies only to the share of net eligible income produced by the modified nexus ratio in article 50ter, alinéa 6 L.I.R. Our IP Box guide works through what qualifies.
Before you fill in a form
The order below front-loads the decisions that are expensive to reverse once a file is open.
- Draw the project boundary in writing first. Set out which costs belong to this project and which are business as usual, because every rate above is a share of eligible costs and the boundary decides what that share applies to.
- Build a sheet for each candidate instrument, carrying its rate, its cap, its eligibility test and the date you checked it — rates differ by company size and project category, not only by scheme.
- Confirm which size column applies under the definition the scheme uses, before any figure derived from that column enters a plan.
- Ask the interaction question explicitly, in writing, for every pair of instruments you intend to combine, and take the answer from the scheme rather than from inference.
- Name who funds the remainder — own funds, bank, SNCI, investors — because each percentage above describes only its own part.
By the time those are settled the remaining work is documentary: a cost breakdown that maps onto the scheme's own categories, a project boundary that holds up when it is re-read, and the annexes each instrument names. Assembling the application file and its annexes is a different skill from choosing between instruments, and it is the half where a well-argued project most often loses time.