Crossroads, Not Cohorts

Why the debate about New Zealand’s startup infrastructure keeps misfiring

The Catalyst

A short exchange ran on LinkedIn this week that captured, in miniature, something I’ve been turning over for a long time. Lane Litz, founder of What Founders Want, posted an observation:

“There are no incubators currently running in New Zealand. 🤯”

By “incubator”, she didn’t mean the equity-taking, cohort-based programmes that have come and gone here over the last fifteen years. She meant the older, looser model: a long-term, community-based environment where founders work at their own pace, with ongoing access to advisors, no fixed cohort, no demo day, and no equity required. A year is more realistic than twelve weeks, she argued, and what founders most need is “community while getting shit done.”

Rowan Simpson responded with the comment that drew most of the attention:

Perhaps the thing you’re missing is that none of these programs are designed for founders. They are really designed for the people who run them. In a hospital, having no sick babies in an incubator would be a milestone to be celebrated, no?

And a second:

We learned that shared working spaces and underwhelming mentors do not attract the best founders — doesn’t matter how many foosball tables and bean bags you have.

An unusually candid pair of observations. Sam Minnée, reflecting on his SilverStripe days, was more nuanced — the looser incubator structure was “less likely to get in founders’ way” than the cohort accelerators that came later, and he valued the community aspect — but landed on a line that gets to the heart of it:

There’s a necessary humility in how much the programme itself is doing — it’s a place where like-minded people can be near each other (including potential advisors); what they do with it themselves is up to them.

Aaron Scott then asked the obvious next question: what does a founder-focused incubator actually look like?

That’s the question I want to take seriously — not because incubators are the right answer, but because answering it properly requires saying something about what New Zealand’s innovation ecosystem actually lacks, and why we keep having the wrong debate about it.

Three things people mean by “incubator”

The word does too much work. An accelerator is a fixed-cohort, time-boxed programme: eight to fourteen weeks, demo day, equity for cash. Y Combinator is the archetype. Most New Zealand programmes that came and went between 2012 and 2022 — Lightning Lab, Creative HQ’s accelerator, Vodafone xone — sat here. A classical incubator, in Lane’s sense, is open-ended, place-based, light on programme structure, often non-dilutive. This is what she’s asking after, and it has largely disappeared from New Zealand. A tech-transfer incubator is a third creature again, existing to commercialise university and CRI research, taking equity in exchange for repayable grant capital. The Callaghan Technology Incubator scheme — WNT, Bridgewest, Sprout — sits here, though its future is now uncertain.

Conflating these three is half the reason the debate goes in circles. Lane is asking why we don’t have the second. Rowan is critiquing what the first largely became. The third is doing its own quite different job.

What the evidence actually says

The empirical picture of accelerators is more sober than the marketing suggests. Careful work — Gonzalez-Uribe and Leatherbee on Start-Up Chile, Hallen, Cohen and Bingham on the broader population — finds real treatment effects, but modest ones. The mechanisms that matter are compressed feedback, network access, and signalling to capital. Most of the apparent magic of Y Combinator is selection: the best founders apply, get in, and would mostly have succeeded anywhere. Rowan’s line — that shared working spaces and underwhelming mentors do not attract the best founders — is empirically defensible.

Notice what’s underneath that finding. Two of the three mechanisms that do work — network access and capital signalling — are not really about programme structure. They are about proximity to the right people. The programme is a delivery wrapper. The active ingredient is the network.

The diagnosis no one quite states

What founders in New Zealand most reliably lack is not a curriculum, a twelve-week sprint, or a building with a foosball table. It is sustained, serendipitous contact with the other people who make innovation happen. Quantum physicists who never meet venture capitalists. Policy thinkers who never sit across from agritech founders. Researchers whose work could change an industry but who have no path to the entrepreneur who could carry it. The default state, in a small and geographically dispersed country, is isolation.

Innovation is a collaborative contact sport. The pattern that produces it — university towns, coffee houses, the early Royal Society, the Wagon Wheel bar in Mountain View where semiconductor engineers swapped salaries and stories — has always been about places where paths cross. Silicon Valley was not built by programmes. It was built by a permeable boundary between Stanford and industry, an open-network norm, and the geographic accident that put a critical mass of the right people inside a thirty-mile radius. The programmes came later and largely codified what the community had already produced.

This is why Rowan’s critique lands so hard. Programmes designed for their operators do not produce connective tissue; they produce throughput. Sam’s defence of the community aspect of the older incubators points at the same thing from the other direction. What was valuable was not the curriculum. It was the proximity, and the trust that built up between people who kept showing up to the same rooms.

Aotearoa’s structural problem is that the rooms have mostly stopped existing. The institutions that historically did this work have been hollowed out, repurposed, or designed for a different job. What remains is a set of programmes optimised for processing cohorts and a deep-tech scheme doing important but narrow work. The connective tissue has thinned.

What the alternative looks like

I should be careful here. I have spent the last several years working on a concept — currently iterating under the working name The Crossroads Collaboratory — that tries to instantiate one version of the answer. I want to be honest about that interest without overselling the project, because what matters is the design problem it gestures at.

The hardest part of getting something like this off the ground in New Zealand has not been money, space, or founder interest. It has been the absence of reference models the country can readily picture. When I describe a salon-style innovator community — diverse by design, anchored in trust rather than contract, measured in years rather than quarters — people pattern-match to the thing they know: a coworking space, an accelerator, an industry association. None of those is the right comparison. The closer reference points are Shack15 in San Francisco, the Long Now Foundation’s gatherings, the old Homebrew Computer Club, and the Royal Institution in its original form. These are not programmes. They are infrastructures of contact, and they produce disproportionate output precisely because they do not try to do what programmes do.

The design principles are straightforward. Place-based but not space-driven: who is in the room matters more than the WiFi password. Curated for ecosystem mix rather than cohort homogeneity — founders, investors, researchers, policy thinkers, storytellers, artists — because the serendipity premium is highest when the room is heterogeneous. Designed for serendipity rather than throughput: salon dinners, fireside conversations, briefings, showcases. Funded so operator incentives align with founder outcomes — not equity, not founder fees, but anchor membership, philanthropic underwriting, or sub-sovereign support that lets operators be measured by what the community produces over a decade. And honest about scope: this infrastructure does not replace capital, IP support, or technical mentorship. It does the one thing those cannot do for themselves. It puts the right people in the same room, repeatedly, over years.

The deep-tech qualification

Deep tech, where my own fund does most of its work, needs more than community: long-horizon translational capital, specialist facilities, regulatory navigation, and technical mentors who have shipped real hardware or biotech. But even there, the deal flow depends on being inside the connective tissue, not outside it. Cyclotron Road and Activate at LBNL are nominally programmes; they are really embedded communities with capital and labs attached. Community-plus, not community-or.

Back to Aaron’s question

What does a founder-focused incubator look like? The honest answer is that the question is slightly misframed. What New Zealand most needs is not another programme. It is the infrastructure of contact that programmes, at their best, were always a partial substitute for. Less like a cohort and more like a club. Less like a curriculum and more like a calendar of high-trust gatherings. Less like a twelve-week sprint and more like a decade-long bet on a thicker network.

The Lane/Rowan exchange is a small public version of a much larger national conversation we keep not quite having. New Zealand’s binding constraint is not founders, ideas, or capital. It is the institutional scaffolding that turns research into export-capable companies, and that scaffolding depends on connective tissue that takes decades to build and minutes to damage. The disestablishment of Callaghan Innovation is a moment of risk for exactly this reason: whatever replaces it will either invest in connective tissue, or it will not.

The right question is not whether incubators went out of style. It is whether we are willing to fund the infrastructure of contact, an innovation ecosystem actually requires, and to measure it by the patience it demands rather than the cohorts it processes.