Shane Solly, portfolio manager with Harbour Asset Management
“The drag on the power grid, locations set back from residential areas, noise, water use and visual effects, many of the centres are industrial-looking – these are the considerations by local communities.
“And the big issue for investors is the timing of the user contracts when these multimillion-dollar assets are built,” says Solly.
“The demand is starting to accelerate for GPU compute and where they get housed. Data centre development is at an interesting junction globally, especially regarding managing the supply of power.
“We have a grid that can supply electricity now and we can provide a degree of firming power to add to the baseload. If there is not enough volume of power or a dry period, a compute load can be moved elsewhere,” says Solly.
Boyes says power purchase agreements (bringing your own power) with hyperscalers are key to developing data centres.
The energy companies can bring incremental renewable generation onstream and the data centre operators and their customers pay for the infrastructure connection. “That’s already happening today.”
Asked what one decision the next government should make to ensure New Zealand captures more of the AI and data centre opportunity, Boyes says: “Establish clear national rules, rather than regional ones, around the issues communities care most about.
“Australia is moving in this direction. Aligning with those frameworks, where possible, will provide greater certainty for investors while helping communities understand how these developments should be managed.”
Boyes says putting rules in place was important so people don’t have to worry, and following the Australian template isn’t a bad start. “Having totally different rules doesn’t make sense.”
Australia is responding to the rapid growth of AI data centres by developing a national regulatory framework to curb environmental impacts, protect local resources, and ensure economic benefits stay within the country.
Prime Minister Anthony Albanese announced plans to introduce binding Australian Standards for AI into Parliament early next year.
The Australian plans also include:
· Energy and Water Rules: Large data centres will face mandatory rules to minimise and report water usage (including prioritising recycled water), cover their own grid connection and network upgrade costs, and ensure power demands do not spike local electricity prices.
· Sovereign Compute and Access: Federal officials have urged operators to provide local start-ups, researchers, and not-for-profit organisations with fair-term access to compute capacity so Australia does not become dependent on imported artificial intelligence.
Boyes says the design of data centres was crucial. “Some designs evaporate water, some have black water, you can’t do that. CDC Data Centres has developed a closed-loop cooling system, and you are good to go.”
Infratil has a near 50% shareholding in CDC, which has three general data centres in Auckland – the first two “are pretty well full to capacity.”
New Zealand hosts about 64 general and cloud data centres (including facilities by providers such as Microsoft, Amazon Web Services, and Datacom), but at present there are no centres built exclusively as a dedicated “AI factory”.
Datagrid NZ has received resource consent to build the country’s first $3.5m AI data centre at Makarewa near Invercargill and is moving through full funding and grid approval. The centre is expected to draw up to 280MW of power.
Mercury Energy is investing $53m for a 12.7% shareholding in Datagrid Holding Group NZ and has signed a 140MW power purchase option for renewable energy.
Contact Energy and CDC are working together on a resource consent for a potential 250MW AI data centre at Stratford near the site of Contact’s Taranaki Combined Cycle gas plant, which has been retired but is close to the transmission grid.
The Stratford facility would be supported by battery storage development. CDC said it would explore a net zero-waste cooling situation, including rainwater collection.
Solly says, “We do have a temperate climate and that’s beneficial from a cooling point of view. That’s a good start to drive less electricity to cool the facilities.”
Boyes says large-scale AI facilities are most efficiently financed by global technology companies with strong investment-grade balance sheets.
At the same time, it can be cost-effective to make additional capacity available within those facilities to support New Zealand’s own AI capabilities and critical services.
“Done well, that has the potential to lift productivity across the economy,” says Boyes.