The label on the topic page decides your budget. Under the Horizon Europe work programme for 2026–2027, the maximum funding rate is 100% of eligible costs for a research and innovation action and 70% for an innovation action, and the same annexes add that other funding rates may be set out in the specific call or topic conditions.
Topics can read almost identically, carry different labels, and leave very different holes in your cash plan. Everything else about joining a consortium — who you talk to, what you own, what sinks the file — follows from reading that label before you agree your share.
The action type decides the rate
A Horizon Europe grant is a rate applied to your eligible costs, not a cheque for a headline figure. The rate is published against the type of action, and the topic text names that type. Founders discover the difference at the worst moment: the coordinator circulates the consolidated budget late in the drafting, and the partner who assumed full reimbursement finds a gap it can now only explain to its board, not renegotiate.
| Action type on the topic page | Maximum funding rate, work programme 2026–2027 | Applied to | What your financial plan must then show |
|---|---|---|---|
| Research and innovation action (RIA) | 100% | your work package's eligible costs | at the maximum rate, no co-funding share is left to source |
| Innovation action (IA) | 70% | your work package's eligible costs | a named, credible source for the share the rate does not reimburse |
Topics also carry a further label — the coordination and support action, for networking, studies and coordination rather than research itself.
Its rate is 100 — but stated as the MAXIMUM Horizon Europe funding rate for a coordination and support action, not the rate as such. The rate actually applied is the one fixed in the specific call/topic conditions and in the grant agreement, and the same page adds that "Other funding rates may be set out in the specific call/topic conditions." Replace the supporting quote with the full list lead-in: "The maximum Horizon Europe funding rates are as follows: ... – Coordination and support action: 100%" (Work Programme 2026-2027, Part 15 General Annexes, p. 33)..
And treat every figure above as a ceiling rather than a promise: other funding rates may be set out in the specific call or topic conditions, which means the rate that actually binds you is the one written into the call you apply under and then into the grant agreement you sign.
Collaborative call or EIC Accelerator
The EIC Accelerator dominates the conversation at startup events, but it is a single instrument inside Horizon Europe and it asks something completely different of a company. The collaborative route asks you to be a competent partner in someone else's research goal. The Accelerator asks you to be the most fundable innovator in a European field. Different file, different calendar, different risk to the cap table.
| Question | Collaborative Horizon Europe topic | EIC Accelerator |
|---|---|---|
| Who the applicant is | a consortium; you own a work package inside it | your company, on its own file |
| What is published as the money | a funding rate — 100% for a research and innovation action, 70% for an innovation action | a grant ceiling of €2,500,000, plus a separate investment component of €1,000,000 to €10,000,000 |
| What it does to your shareholding | nothing: the grant reimburses costs you have incurred | blended finance means a grant component plus direct equity or quasi-equity such as convertible loans |
| Time box on the funded work | the work plan the consortium submits | the grant component runs to a maximum of 24 months |
Scale sharpens the contrast. For the 2026 call the EIC published €220,000,000 behind its Challenges strand and €414,000,000 behind its Open strand, and the Accelerator is not a well you can return to at will: 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.
A collaborative call, by contrast, is a recurring channel — the same partners reappear across topics for years, which is why relationships matter more than any single submission.
What the assessors are actually reading
The proposal a first-time partner imagines is a document about technology. The proposal being assessed is a document about delivery. Your section is short, it is read by people who have read many variants of it, and it is judged on whether a stranger could run your work package from what you wrote. The failures below recur, and none is about the quality of the science.
| Part of the file | What is genuinely being read there | Where a first-time partner loses it |
|---|---|---|
| The administrative screen | whether every named participant is registered, validated and documented | a partner whose registration paperwork is still unfinished in the final week |
| The science section | method, state of the art, and what is actually new | a work package written in marketing language, which the research partners cannot defend for you |
| The impact section | who exploits each result, and how | no exploitation claim of your own, so you read as a subcontractor with a logo |
| The implementation section | the work plan, the effort split, the budget | effort that does not reconcile with the money, or an uncovered cost share with no source named |
The last row does the quietest damage. A budget whose uncovered share has no origin describes a project that cannot be delivered as written, and an assessor does not need to argue with your technology to mark it down.
How Luxembourg companies actually get into a consortium
Almost nobody's first project starts with writing a proposal alone. The realistic entry paths, in the order they tend to work:
- Join, do not lead. First-timers enter as partners under experienced coordinators. Your job is a defensible work package and clean paperwork, not the architecture of the whole project.
- Use the national support structure early. Luxinnovation is the national agency that handles European funding advice, and the conversation is worth having before you commit to a topic, not after.
- Be findable. Consortia are assembled by people searching for a specific capability. A precise public description of what your company can technically do is bait; positioning language is not.
- Read the projects that were already funded. CORDIS carries the Commission's record of funded projects — where you find the coordinators already working in your area, and the consortia that will need a partner like you next time.
- Start from your own R&D record. A company already running nationally funded R&D arrives with exactly what coordinators want: demonstrated capability delivered under public-money discipline.
Calls themselves live on the EU Funding & Tenders Portal, which is also where a topic's own conditions live: each topic sets its own conditions, including what it expects of a consortium, and those conditions are checked before anyone reads your science.
There is a quieter route too: some funded consortia redistribute part of their budget to third parties through open calls of their own, under conditions and ceilings set by the programme and restated in each project's call text. Those files are a fraction of the size of a full proposal.
The stacking question
National and EU instruments are built to sit at different layers, and a Luxembourg SME with real R&D can usually populate more than one. Aid for R&D projects covers, at the base rate for a small enterprise, up to 60% of industrial research costs and up to 40% of experimental development costs. The young innovative enterprise scheme goes up to €1,000,000 at a maximum 70% co-funding rate for companies up to 5 years old.
Further downstream, the IP Box applies only where the resulting asset is one article 50ter L.I.R. covers — patents, utility models and self-developed copyrighted software, not marketing intangibles — and then only to the share 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.
Whether a specific national aid may sit on the same cost line as an EU grant is a cumulation question, and it is one to settle with Luxinnovation before the budget is fixed rather than after the file is submitted. Getting that sequence backwards is how a partner ends up rewriting a budget in the week it should be proofreading one.
The questions that come next
These come up once the strategy stops being abstract and a specific topic is open on the screen.
Can a company formed this year hold a work package?
The published funding rate attaches to the action type, not to the applicant's age. What a call requires of its participants is written in that call's own conditions, and the practical gate is separate again: a coordinator has to be willing to carry an unproven partner through an assessment. Evidence beats age, which is why a nationally funded R&D project on the record is worth more than another year of simply existing.
Do we need a previous EU project to be taken seriously?
No coordinator asks for one as such. What they are testing is whether you can deliver a work package on time and account for public money afterwards, and a completed national R&D project answers that question with documents rather than assertions. It is the cheapest credential available to a Luxembourg SME, and the reason national instruments belong before European ones in most plans.
What does owning a work package actually cost us?
Real senior time, and not only during the proposal: it recurs for the life of the project. You take on named deliverables with dates, effort you have to record and report, and a share of the periodic reporting the coordinator assembles. Project timelines move at consortium speed, so a company whose survival depends on this quarter should not anchor its plans to a Horizon call.
Is the SME Instrument still a separate programme?
No — the SME Instrument was incorporated into the European Innovation Council when its pilot phase launched in 2018, together with the Future & Emerging Technologies programme. Advice written around the old brand predates that change, so check the date before following it.
The sequence that works
Most of a proposal calendar disappears into the unglamorous work of assembling the budget behind your work package and the annexes that have to support it, while the science section gets written last, by people who already know what they want to say. Run it in this order:
- Read the topic's label and conditions before you agree anything with a coordinator; the action type is the rate you will live with for years.
- Finish the registration and validation paperwork a coordinator asks for as soon as you have a topic: it blocks nothing until the final week, when it blocks everything.
- Write your co-funding line first. Name where the share the rate does not reimburse comes from, and put it in the plan rather than in your head.
- Take the cumulation question to Luxinnovation before the budget is locked.
- Ask the coordinator for the effort table early, and reconcile your person-months against your budget yourself.
- Have someone outside the project read your work package and find its weakest sentence, because the assessment will.
- Plan capacity for a second cycle. A partner who treats one submission as the whole strategy tends to write like it.