Horizon Europe has signed 6,620 collaborative grants since 2022, worth 35.7 billion euro, with 23,843 organisations in the consortia behind them. Read the participation rules and the programme looks wide open: any legal entity, any member state, three partners minimum. Read the funded record and it looks different. Most consortia are reunions, and most organisations never get a second invitation.
Every figure below comes from the funded projects themselves - who sat in which consortium, in what role, and for how much - as of October 2026. Nothing is surveyed or estimated.
- 1 in 4 - partner pairs in 2026 consortia that had already won together under Horizon Europe
- 35% - share of all collaborative funding held by the top 1% of organisations
- 62% - share of organisations that appear in exactly one project
- 5% - share of university seats in 2024-2026 projects taken by universities new to the programme
A few hundred organisations, a third of the money
Horizon Europe's collaborative record has a thin head and a very long tail. Of 23,843 organisations that have held a coordinator or participant seat, 14,793 have held exactly one. At the other end, 171 organisations appear in more than fifty projects each, and between them they hold 22% of all seats.
Money is more concentrated than seats, because the most active organisations also take the larger budgets.
| Organisations ranked by activity | Share of seats | Share of funding |
|---|---|---|
| Top 10 | 4.2% | 6.9% |
| Top 1% (238 organisations) | 26.0% | 35.2% |
| Top 5% (1,192) | 48.1% | 57.4% |
| Top 10% (2,384) | 59.3% | 68.6% |
| Top 25% (5,960) | 74.5% | 81.2% |
The top of the list is not a surprise: Fraunhofer leads with 628 projects and 526 million euro, followed by CERTH in Greece, CNR in Italy, CSIC in Spain, CEA and CNRS in France. What matters more than the names is how far their reach extends. In every year since 2022, between 79% and 88% of funded consortia have included at least one of the 171 most active organisations. Four in five projects are coordinated by an organisation in the top tenth of the activity ranking, and between 39% and 47% by one in the top hundredth.
Coordination is the most concentrated role of all. Of 2,362 organisations that have ever coordinated a collaborative project, 1,442 did so exactly once. The top 1% of coordinators account for 15% of all coordinations; the top 10% for half.
Consortia are reunions
Concentration says who shows up. It does not say whether they show up together. For that, every funded consortium was broken into its partner pairs - 697,370 pairs across all projects - and each pair was checked against the record of earlier-signed projects.
In projects signed in 2026, 25.5% of partner pairs had already won a Horizon Europe grant together. That is a lower bound: the record starts in 2022, so partnerships formed under Horizon 2020 and earlier are invisible here and count as first meetings.
The coordinator's own network is tighter still. In 2026, 40% of participant seats were filled by organisations the coordinator had already won with under Horizon Europe. In 81% of consortia the coordinator had prior history with at least one partner, and in 37% with at least half of them. Only 19% of coordinators assembled a consortium with no prior Horizon Europe history at all.
One thing this data does not show is a trend. The cumulative repeat share rises every year - 9% in 2022, 19% in 2024, 26% in 2026 - but that is bookkeeping: each year adds history for the next to match against. Measured against a fixed two-year window the share moves the other way, because the window itself shrinks as the programme matures. Neither measure is clean. What is robust is the level: roughly a quarter of pairs and two fifths of coordinator-partner ties are repeats, and there is no evidence the network is closing further.
Across 6,282 pairs that have won together five or more times, only 12.7% share a country - barely above the 11.5% of all pairs. The strongest ties in Horizon Europe are European, not national. The most frequent partnership on record is Fraunhofer and CEA, together in 91 funded projects.
| Partnership | Projects together |
|---|---|
| Fraunhofer (DE) and CEA (FR) | 91 |
| VTT (FI) and Fraunhofer (DE) | 67 |
| CNR (IT) and CNRS (FR) | 64 |
| imec (BE) and Fraunhofer (DE) | 55 |
| TU Delft (NL) and Fraunhofer (DE) | 54 |
| TNO (NL) and Fraunhofer (DE) | 52 |
| Wageningen Research (NL) and Wageningen University (NL) | 52 |
| Tecnalia (ES) and Fraunhofer (DE) | 48 |
| CEA (FR) and CNRS (FR) | 48 |
| VTT (FI) and TNO (NL) | 45 |
Where the door is narrowest
Repeat partnership is not evenly spread. Some parts of the programme are near-closed circuits; others are genuinely open. The share of partner pairs with prior Horizon Europe history, for projects signed 2024 to 2026:
Research infrastructures and air-traffic research are clubs: nearly half of every consortium has worked together before. Cluster 2 and the bio-based Joint Undertaking are the most open parts of the programme, with repeat shares a third of that.
The grant format matters as much as the field. In 2026, coordination and support actions had a 36% repeat share, research actions 27%, and innovation actions 15%. Innovation actions bring in companies and demonstration sites that the research core has not worked with, which keeps them open. Size cuts the same way: consortia of six or fewer partners had a 32% repeat share in 2024-2026, against 21% for consortia of more than fifteen. Small consortia are the ones formed by phone call.
Who gets in as a newcomer
A newcomer here is an organisation holding its first Horizon Europe collaborative seat. In projects signed 2024 to 2026, newcomers took 20 to 24% of seats and 16 to 17% of the money. They coordinated between 6% and 12% of projects, depending on the year.
Which newcomers get in depends almost entirely on what kind of organisation they are.
| Organisation type, 2024-2026 seats | Share taken by newcomers | Share of all newcomer seats |
|---|---|---|
| Universities | 5.4% | 6.3% |
| Research organisations | 8.0% | 8.7% |
| Public bodies | 33.4% | 8.7% |
| Companies | 38.8% | 58.8% |
| Other (associations, NGOs) | 34.6% | 17.5% |
The academic core of the programme is essentially settled: 19 in 20 university seats go to universities already in the system. Companies are the revolving door. Nearly two in five company seats go to companies new to the programme, and companies account for six in ten of all newcomer seats.
Newcomers also get less per seat. A newcomer company receives 367 thousand euro on average against 436 thousand for a returning one; a newcomer research organisation 386 thousand against 633 thousand. The first seat is usually a small work package, not a leading role.
The revolving door also turns the other way. Of organisations that entered in 2022, 59% have appeared in at least one further project since - but 82% of the universities and research organisations against 49% of the companies. For the 2023 cohort, with less time elapsed, the figures are 50% and 28%.
Half of the companies that win a first Horizon Europe grant never win another. That is the figure that should worry an SME planning its EU funding strategy around one project.
The on-ramp is other newcomers
Where do newcomers actually find their seat? Not, for the most part, with the hubs. Consortia coordinated by an organisation with twenty or more prior projects gave 22% of participant seats to newcomers in 2024-2026. Consortia coordinated by a first-time coordinator gave 36%.
| Coordinator's prior Horizon Europe projects | Participant seats | Share to newcomers |
|---|---|---|
| None (first-time coordinator) | 1,562 | 36.3% |
| 1 to 4 | 3,964 | 24.9% |
| 5 to 19 | 8,630 | 22.4% |
| 20 or more | 19,009 | 22.2% |
Experienced coordinators do seat newcomers - 22% is not nothing, and they run most of the projects - but their default is the known network. The consortia with the most open doors are the ones built by coordinators who do not yet have a network to fall back on.
Geography follows the same pattern as organisation type. Seven countries hold 64% of the money: Germany, Spain, France, Italy, the Netherlands, Belgium and Greece. The fifteen countries covered by the Widening measures hold 22% of seats and 18% of funding. Greece is the outlier among them, with 5.7% of funding but 8.6% of all coordinations and two of the fifteen most active organisations in Europe.
What this means for a bid
The published record changes how a consortium should be built, especially for an organisation that is not yet in the network.
- Your first seat comes from a coordinator, not a call. Most funded consortia were assembled from prior working relationships. Finding the organisations that coordinate repeatedly in your field, and what they delivered, is the work that precedes any proposal.
- Choose the open parts of the programme. In innovation actions, large consortia, Cluster 2, Health and the bio-based Joint Undertaking, the share of partners who have worked together before is a third to a half of what it is in research infrastructures or small coordination actions.
- A newcomer-led consortium is a real route. First-time coordinators fill a third of their consortium with other newcomers. The success rate is another question, covered in our previous analysis, but the door is wider.
- Plan for the second project at the first. Half of companies never return. The ones that do are those that turned one seat into a relationship with the coordinator, which is what the repeat-pair figures measure.
GrantSignal holds every one of these consortia, with each organisation's track record, its most frequent partners, and the documents its projects produced. Describe what you do and it suggests the organisations most likely to need you - and shows you who they have won with before. The live demo runs on the same database.
What this analysis cannot say
- It cannot see Horizon 2020. Every repeat figure is a lower bound. Partnerships formed before 2022 register here as first meetings, and the true repeat share is higher.
- It cannot measure a trend. Both cumulative and fixed-window repeat shares move for mechanical reasons as the record grows. The level is well supported; the direction is not identifiable from five years of one programme.
- It cannot see rejected proposals. The data covers funded consortia only. Whether repeat consortia win more often than new ones, or simply apply more often, cannot be separated without the unfunded applications.
- Signature years are not call years. Grants from 2025 calls were mostly signed in 2026, so single-year figures for 2025 rest on 572 projects and are the least settled.
- Retention figures are censored. Organisations that entered in 2024 or later have not had time to return; only the 2022 and 2023 cohorts are mature enough to read.
Method, briefly
Built from 6,620 Horizon Europe collaborative projects signed between February 2022 and August 2026, covering research and innovation actions, innovation actions, coordination and support actions, and their Joint Undertaking equivalents. Fellowships, ERC grants and EIC Accelerator grants are excluded because they are not consortium instruments. Only coordinator and participant roles count as seats; associated partners and third parties are excluded.
An organisation is identified by its participant identification code. A partner pair is a repeat if the same two organisations appear together in a project with an earlier signature date. A coordinator "knows" a partner if the two appeared together, in any roles, in an earlier-signed project. A newcomer is an organisation whose first seat was signed in the year under study.
The queries that produce every figure above are published alongside the funding-bar notebook in the GrantSignal repository, and re-running them against current data is how this article is kept accurate.