Whichever coalition forms out of this past weekend’s election has a substantial opportunity ahead of it.
America’s midterm elections brought a Congress that may be more inclined to take back authority over tariffs from the President. Uncertainty over trade will hopefully decline. It would be good for Americans and good for us as well.
The elections also brought another shift that’s been less noticed.
American elections include referenda on local or state-wide initiatives. A series of ballot initiatives sought to restrict new datacentres.
In some counties, substantial taxes paid by datacentres had helped reduce everyone else’s property taxes. Competition for that investment has brought tax concessions that are often opaque. Legitimate worries about concessions combined with other concerns around issues like water-use and energy prices, bringing voter backlash. Voters favoured greater restrictions in 46 of the 47 local datacentre ballot measures tracked by Ballotpedia last week; three decisions were still pending at time of writing.
Water use has been overblown as a concern. Southland’s large datacentre can take, at most, about as much water as a golf course and far less water than a meatworks.
But America’s electricity grid has not been able to keep up with demand. In the longer-term, demand from new American datacentres can help push down power prices by helping to cover the cost of grid upgrades. But new power plants face years of delay before they can connect to the American grid. Power prices can go up in the interim.
Bans based on short-term bottlenecks in the US are misguided. But they bring opportunity for other places, and especially if operators want to hedge their bets about American policy.
Two years ago, I worried that this opportunity could be time-limited. Companies like Anthropic and OpenAI had enormous short-term demand based, at least in part, on hopes of being first to develop a super-intelligence.
Demand for datacentres is unlikely to abate anytime soon.
Even if artificial intelligences do not get smarter than today’s state-of-the-art models, only a tiny proportion of the world’s population currently uses those frontier models. It is legitimately difficult to tell what the world will be like when access to those models is broadly dispersed – a bit like imagining today’s smartphone-world from observing a few people with car-phones in the mid-1980s.
And futures with much smarter intelligences are harder to wrap our heads around.
But a very wide range of plausible futures will need more datacentres and the energy that powers them.
New Zealand has enormous potential for new conventional energy generation and the opportunity to accelerate progress towards new forms of geothermal energy.
Datacentre demand is already helping build new generation. In 2022, Contact and Microsoft signed a ten-year renewable-energy agreement that helped back new geothermal investment. In August 2026, Contact announced another proposed data-centre partnership that would back investment in new geothermal, wind, and solar.
Long-term contracts from datacentres can underpin investment in new electricity generation. It can also fuel research into improved forms of energy generation because access to cheaper and more reliable power generation makes datacentres more cost-effective.
America’s Fervo Energy makes geothermal power viable in places where it previously had not been cost-effective. Its 2023 commercial pilot came under partnership with Google, who wanted more sources of clean energy. In September 2026, Google and Fervo announced a power purchase agreement enabling continued development and giving Google the option to take up more electricity in the future.
New Zealand has ample conventional geothermal potential that can be developed, if demand is there for the electricity – alongside other generation options. But it also has an opportunity that is harder to develop elsewhere. Very hot rocks deep underground contain far more energy than conventional geothermal. And Taupo volcanic zone’s very hot rocks are only 4-5 kilometres below ground level in some places; elsewhere they can be much deeper.
The government has put $60 million toward an exploratory well investigating this kind of ‘supercritical’ geothermal potential. They hope to start drilling next year.
Datacentre demand can scale up conventional geothermal energy development. And it could also encourage commercial investment speeding up progress toward ‘supercritical’ geothermal energy. Commercial generation costs for supercritical geothermal are hard to forecast until the test wells progress. But there is potential for cheaper and near-limitless energy.
Both National and Labour have promised new requirements for datacentres’ electricity supply.
But the real constraint is not datacentres’ willingness to partner with power companies to build the generation that will power their sites. Datacentres cannot do much without power. The contracts that secure that power also fund new generation. If an incoming government wants datacentres to bring additional generation, it should make that generation easier to build.
Clearer, faster routes for getting permission to develop geothermal resources, alongside modernisation of the regulations for developing those sites, would help. And especially when building in America is not getting easier.
MBIE promises work in 2028 and 2029 to “ensure geothermal regulatory frameworks are fit-for-purpose and for next-generation geothermal technologies.”
An incoming government, whether National- or Labour-led, should take advantage of the current opportunity by accelerating that work.
Read the article on The Post here