Social Clusters
Mixed cohorts and shared-service operators, with an anchor
LATAM has run programs that built networks and then let them dissolve when the cohort ended. Run curated mixed cohorts with retention incentives, tracking and an institutional anchor funded from day one.
The rationale
Scientific talent does not convert into companies without connection infrastructure. A 2018 study of Chilean biotech firms found almost half of business links, 47%, originated in personal connections, and analyses of global hubs describe individual brokers linked to as many as 15 ventures at once. Where those brokers are scarce, a founder pays in diligence time, in the search for a first pilot customer, and in capital that flows to whoever is already known. The region has funded labs, grants and equipment far more consistently than it has funded relationships, and the 2024 investment split shows the result: Chile drew US$607m of private deep tech investment against Brazil’s US$216m.
Two models are worth copying, with their conditions attached. Start-Up Chile offered equity-free US$40,000 grants and one-year visas and enforced mixed cohorts of foreign and Chilean teams; more than 3,000 startups from 85 countries passed through it. Colombia’s Alianza DeepTech is the region’s one worked Ecosystem-as-a-Service instance, uniting 31 universities, corporations and government agencies under joint action commitments and a shared KPI set. The move is to replicate the Alianza structure in more countries and to run curated cohorts mixing domestic and international founders, with retention incentives, alumni tracking and a named institutional anchor written into the design before the first cohort is recruited.
The honest objection comes from the chapter’s own evidence. Start-Up Chile imported global know-how successfully, but without retention and institutional anchoring most of the value dissipated once participants left. The peer-learning benefit was also asymmetric: a 2014 study found only 16% of foreign founders named it their primary source of value against 45% of domestic founders. Mixed cohorts are therefore in large part a subsidy to the domestic side, which is defensible when that is the stated objective and the anchoring is funded. Run without an anchor, a program buys twelve months of density and leaves nothing standing.
The evidence
What it unlocks
Cohorts that leave an institution behind, so the introductions, the shared services and the alumni are still there after the program’s funding ends.
It has been done before
The Cambridge cluster and Cambridge Network
Cambridge Network is a membership institution whose only product is convening: introductions, sector groups and recurring events across a cluster of more than 5,000 knowledge-intensive companies that together turn over more than £18bn a year and employ more than 67,000 people. The convening layer has its own funding and its own staff and outlives any single program, which is exactly the anchoring Start-Up Chile lacked.
First moves
- 0–6 months
Fund the anchor before recruiting the cohort
Name and fund the institution that will hold the network after the program ends, a university TTO, a national alliance or an Ecosystem-as-a-Service operator, with a three-year budget line and a named alumni-relations owner, before a single cohort is recruited.
National innovation agencies (CORFO, ANII, iNNpulsa) + university TTOs - 6–18 months
Replicate the Alianza structure in two more countries
Constitute a domestic deep tech alliance of universities, corporates and public agencies with published joint action commitments and a shared KPI set, on the Alianza DeepTech Colombia template, in two additional countries.
National alliances + LADP coalition secretariat - 18–36 months
Run mixed cohorts with retention terms attached
Run curated cohorts mixing domestic and international founders where the grant carries a stay-and-mentor condition, alumni are tracked for 36 months, and the anchor institution publishes where they went and what they built.
Accelerators and venture builders (Start-Up Chile, GRIDX, Ganesha Lab) + the funding agency
How we would know it worked
- Share of cohort alumni still operating in the host country 24 months after the program ends, tracked by the anchor institution, above 50%.
- Domestic deep tech alliances constituted with published joint commitments and a shared KPI set, rising from one (Colombia).
- Business relationships that originated through a program-run introduction rather than a pre-existing personal connection, surveyed annually against the 47% personal-connection baseline.
- Recurring convenings held per country per year and unique organizations attending, published by each anchor institution.
What could go wrong
The cohort ends and the network leaves with it
This is the documented failure mode rather than a hypothetical: Start-Up Chile imported know-how and most of the value dissipated when participants left. Containment: fund the anchor institution on a longer clock than the cohort, attach retention conditions to the grant, and publish alumni location and status every year so dissipation is visible while it is still fixable.
Paying for foreign founders who take the value home
An import-and-mingle program funds visitors whose main benefit accrues to their domestic peers, and only 16% of foreign founders in the 2014 Start-Up Chile study named peer learning their primary source of value. Containment: state the domestic-benefit objective explicitly, size foreign intake to the domestic cohort it is meant to teach, and report outcomes for the domestic side rather than the headline count of nationalities.
Who acts
Rests on
01 Venture Building The science-to-startup pipeline
15 A Pan-LATAM Forum One flagship, one annual KPI release
03 Showcase & Visibility Educate privately, convene publicly
The data behind it
Brazil hosts 41% of the region’s ventures, more than the next four countries combined. The “Big 5” account for 83% of everything mapped.
As published in the report, September 2025.
SourcesLatin American Dynamism Project (LADP), Accelerating Deep Tech in Latin America; Tracxn Technologies, Tracxn, LATAM deep tech company database (LADP extract); EMERGE (Emerge Brasil) & Cubo Itaú, Radar Deep Tech LATAM 2025
Methodology: LADP/Tracxn count and share. ledger #25/#26: EMERGE's independent mapping disagrees sharply (Brazil 72.3% vs 40.8%; different rank order). Neither is "the" number, a definitional gap, not an error.
| Category | Current |
|---|---|
| Brazil | 1,048 |
| Mexico | 358 |
| Argentina | 256 |
| Chile | 251 |
| Colombia | 219 |
| Rest of LATAM | 434 |