I’m lucky enough to sit at governance tables across infrastructure, technology, health, insurance and manufacturing. Economics, business models and balances of power are moving at once. After a couple of years playing with emerging technology and sitting around tables with people smarter than me, wondering what “good” might look like for this small, isolated, hydro-rich piece of rock, here’s my take. Take it, argue with it, improve it.
Let’s get the uncomfortable bit out of the way. We were never going to build the next great AI brain. That game is being played by a handful of enormous companies with budgets bigger than our GDP, chasing chips that mostly come out of one factory on the other side of the world. If your plan for New Zealand involves outspending those guys, I’d suggest a lie down and a cup of tea. The good news is that the AI arms race is also a colossal money pit. Somebody has to make all that spending pay off eventually. It very much isn’t going to be us. Good. Let them worry about it.
Our opportunity sits somewhere much less glamorous. Somebody has to actually run this stuff: keep it powered, working, secure, and accountable. That work has to happen somewhere, and it doesn’t all have to happen across the Pacific. We don’t need to invent the technology to extract enormous value from it. We need to get good at pointing it at things we already know how to do: farming, health, engineering, manufacturing and our own language. Local teams are already building speech recognition for te reo Māori that works remarkably well, using technology on our own terms. That isn’t a cute little side project. It is a template.
We could also become the people the rest of the world trusts to run AI properly. Much AI is being deployed that nobody can properly explain, audit or hold to account. New Zealand has a functioning legal system, relatively strong institutions and a reputation for doing things properly. In a world drowning in AI nobody can vouch for, being trusted to run it well is potentially worth real money. Then there is human capability. We don’t necessarily need more researchers chasing the bleeding edge. We need farmers, clinicians, engineers and tradies who understand AI as well as their own trade. They can use these tools to become better at what they already do. That is expertise we could build at scale.
There is also a role in shaping rules, and carefully selling some of our abundant clean electricity into the hosting side of this economy. But this is where the debate becomes embarrassingly binary. Every time a big data centre is proposed, we seem to lurch towards either “ban it” or “build it”, as though everything in between represents moral failure. Sensible questions include: who pays for additional generation and network capacity? What happens to power prices? Does the water and land use stack up locally? Who owns the asset? What jobs are actually created? And did the community get a meaningful say before the concrete went in?
Those are legitimate trade-offs, and reasonable people can land in different places. Show me a significant economic decision without trade-offs and I’ll show you an argument that probably hasn’t been thought through. The interesting conversation is in the middle. Not blindly waving everything through because someone promised jobs, and not reaching for a ban because the issue is complicated. There is another option: doing nothing. Buying whatever technology is offered, handing over the decisions along with the money, and hoping for the best. More than half of New Zealand’s small businesses currently have no plan to meaningfully engage with AI. That isn’t necessarily caution. At some point, it becomes drift.
And here is the line I’d actually like people to remember. It is perfectly sensible to buy chips and clever software from overseas. What we should never outsource is our judgement: who decides how the technology is used, who controls the data, who can access it, and what happens when the supplier changes the rules, gets bought, goes broke or decides we’re no longer an interesting customer. Independence was never about making everything ourselves. It is about retaining the ability to choose.
That means every organisation in the country, not just government and the technology sector, should ask some basic questions. Where does our data actually live? Under whose laws does it sit? Who can be compelled to hand it over? What happens if the service disappears tomorrow? And the question almost nobody asks: have we actually tested any of this, or did we just write it into a policy and hope?
We’re a small, soggy, hydro-blessed democracy clinging to the edge of every map, connected to the rest of the world by a handful of cables under the sea. There is a genuinely good version of the AI story for New Zealand. It doesn’t involve pretending we can beat Silicon Valley at its own game, nor hiding under the bed until the whole thing goes away. It involves being smart about where we play. Quietly, competently and with a bit of bloody-minded independence, we could become very good at being the people others trust to use, manage and govern this stuff.
Nobody is going to build that foundation for us. The clever bit upstairs is being built regardless of what we do. The groundwork underneath it is still entirely up to us. So perhaps we should get on with it.

Thanks Ben. A thoughtful distillation of how we can we leverage AI in our distinctive kiwi way to improve what we do well and become a trusted user and manager of AI.
It’s a good article and I agree with the sentiment. But new Zealand is also full of pc political speak. I work in govt and have been working on trying to apply ai. The real roi is redesigning workflows not automating archaic ones. And govt is number one sector for roi however the mandate and support and knowledge barely exists. I have been working on auditable governance systems and systems to change workflows. It doesn’t matter me popular. In fact my jobs often on the line. The human factors and leadership is huge issue. Good leaders r often the ones tinkering away and the ones who want to be leaders are often not the best for job. This is going to matter a hell of a lot more with AI with decisions and systems in seconds instead of months