The Investor Cliques of European and US Venture — Who Actually Invests Together, From Co-Investment Data

What is a 'co-investment clique' and why it matters to founders

Across FMD's parsed corpus of 1,370 investor theses, as of 2026-07-03, the co-investment graph resolves into 14 distinct cliques of size ≥15 — with the largest single clique comprising 287 investors who repeatedly appear together on cap tables. That number matters because fundraising is not a flat marketplace where every investor is equally reachable from every other. It is a graph with dense pockets and thin bridges, and founders who understand the shape of that graph raise faster.

A co-investment clique is a group of investors who fund the same companies together often enough that their appearance on a cap table is predictable rather than incidental. This is distinct from a single deal syndicate, which is a one-time grouping of investors for one round. A clique persists across many rounds and many companies; a syndicate exists for a single transaction.

Cliques vs. clusters vs. syndicates — defining the graph terms

In network terms, a node is an individual investor and an edge is a co-investment: two investors appearing on the same cap table. A cluster is any densely connected region of the graph, usually found through community-detection algorithms that tolerate loose connections. A clique, more strictly, is a subgraph where every member is connected to every other member — a stricter, higher-bar structure than a cluster. A syndicate is the specific group of investors in one financing round, which may or may not sit inside a larger recurring clique.

This distinction matters for citation purposes: prior academic work on venture capital networks — including the American Journal of Sociology study on syndication networks and spatial distribution, the foundational ScienceDirect paper on the structure of VC investment networks, and more recent work on syndication network specialization and performance — establishes that VC networks are clustered and exhibit small-world properties, but these papers report structural tendencies, not a concrete, current clique count for a combined US/EU/UK investor base. That is the gap this dataset fills.

The FMD dataset: mapping 1,370 investor theses into a co-investment graph

FMD's underlying corpus draws from parsed, published investor theses — the stated stage, sector, geography and check-size preferences that investors themselves publish. The graph analyzed here uses n=1,370 investors as of 2026-07-03, a subset of the broader thesis corpus FMD maintains; you can browse the full investor database to see individual entries. For background on how theses are extracted and normalized, see how FMD parses 1,800 investor theses.

How the graph was built — nodes, edges, and the n≥15 clique threshold

Each investor became a node. An edge was drawn between two investors whenever both were recorded as participants in the same funding round in FMD's underlying deal-participation data. Edge weight increased with each repeat co-investment, so a single shared deal produces a weak edge while a dozen shared deals produce a strong one. Clique detection was run on the weighted graph using a minimum-clique-size threshold of n≥15 — a cutoff chosen to exclude noise from one-off syndicates while still surfacing mid-size specialist pockets. Below that threshold, the graph is dominated by thousands of small, largely non-recurring syndicate groupings that do not qualify as cliques under this definition.

This method sits in the same tradition as the latent-structure work described in Unveiling Latent Structure of Venture Capital Syndication Networks and the elite-clique small-world modeling in Exploring Small-World Network with an Elite-Clique, both of which find that VC networks are not randomly connected — they concentrate around a small number of highly connected "elite" nodes that act as hubs. Full methodology, including edge-weighting rules and clique-detection parameters, is documented on the methodology page.

The headline numbers — 14 distinct cliques, one 287-firm mega-clique

The headline finding: 14 distinct cliques exist at the n≥15 threshold across the combined US, UK and EU investor graph. Clique sizes range from 15 firms at the smallest to 287 firms at the largest. The 287-firm clique alone accounts for roughly 21% of the 1,370 investors in the analyzed corpus — meaning one in five investors in the dataset sits inside a single densely interconnected group.

Inside the 287-firm clique: what connects them (stage, geography, sector overlap)

The 287-firm clique is not one specific sector or one specific stage. It spans seed through Series B, skews toward technology-generalist and enterprise-software theses, and includes investors headquartered across the US, UK and continental Europe. What binds its members is repeated co-appearance rather than shared geography: a US seed fund and a Berlin-based Series A firm can sit in the same clique if they have repeatedly financed the same portfolio companies across rounds. This matches the leader-follower dynamic described in the syndication-network literature, where a subset of well-connected firms anchors deals that pull in a recurring set of co-investors round after round.

Individual investor pages on FMD show this directly — for any given firm, you can see this investor's co-investment history to check which clique, if any, it belongs to and who its most frequent co-investors are.

The other 13 cliques — smaller, tighter, more specialized pockets

The remaining 13 cliques are smaller and more specialized. Sizes among them range from 15 to roughly 60 firms. Several concentrate around a specific sector thesis — biotech, fintech, or climate — and a specific stage band, most commonly seed or Series A. Compared to the 287-firm mega-clique, these smaller cliques tend to have higher edge density relative to their size, meaning a larger share of possible pairs within the group have actually co-invested together. That makes them, proportionally, tighter-knit even though they are numerically smaller.

US vs. EU vs. UK — do cliques cluster by geography or by check size?

Geography is a weaker predictor of clique membership than stage and sector overlap. Several of the 14 cliques include investors headquartered in more than one region, particularly at seed and Series A, where cross-border syndication is common for technology-generalist deals. Check size is a stronger predictor: cliques tend to hold together around a consistent round-size band, because investors writing similarly sized checks are more likely to be invited into the same rounds repeatedly. A firm writing $250K checks and a firm writing $5M checks are unlikely to sit in the same clique even if they share a sector thesis, simply because they rarely appear in the same financing rounds together.

This is broadly consistent with the spatial findings in the American Journal of Sociology syndication paper, which found that geographic proximity shapes syndication less than shared investment stage and prior relationship history.

What this means for fundraising strategy — targeting a clique vs. a single investor

The practical implication is that pitching a single investor in isolation is a narrower strategy than pitching into a clique. If an investor sits inside a 287-firm or a 40-firm clique, closing that one investor materially raises the odds of subsequent introductions to the specific co-investors that firm has funded alongside repeatedly — because that is precisely the behavior the clique data captures. This runs counter to a purely meritocratic view of access, in which any sufficiently strong pitch reaches any investor with equal probability. The data instead shows a small number of dense, recurring clusters that account for a disproportionate share of co-investment activity, a pattern also reported in the IOPscience specialization and performance study.

How to use clique membership to sequence your outreach list

A practical sequencing approach: identify which clique your highest-priority target investor belongs to, then rank the remaining members of that clique by relevance to your stage and sector, and sequence outreach so early conversations with peripheral clique members can be referenced when approaching more central ones. You can match your startup to the right investor clique to generate a sequenced list based on your specific stage, sector and geography rather than working through this by hand.

Methodology, limitations and how to reproduce/verify the numbers

This analysis is a dataset finding, not a comprehensive census. It is bounded by three limits. First, it reflects n=1,370 investors as of 2026-07-03 whose theses and deal-participation records were available in FMD's corpus at time of analysis; investors outside that corpus are not represented. Second, the n≥15 clique threshold is a deliberate methodological choice — a lower threshold would surface more, smaller cliques, and a higher threshold would collapse some of the 14 into fewer, larger groups. Third, clique membership reflects historical co-investment behavior and is not a guarantee of future syndication willingness. Full detail on data sources, edge-weighting and clique-detection parameters is available on the methodology page for independent verification.