MYTHBUSTERS
Why New Zealand Can Become a Global Hub for AI Data Centres
As global demand for artificial intelligence (AI) accelerates, the underpinning infrastructure - data centres, electricity generation, transmission and connectivity - has become one of the most critical and investable asset classes globally.
Following the recent planning approval for Datagrid’s 280 MW Makarewa AI data centre in Southland, a familiar set of myths has resurfaced about whether New Zealand can - or should - play a material role in this industry. The reality, as with New Zealand’s renewable energy prospects, is very different.
“New Zealand does not need to become the world’s largest AI data centre market to win. It simply needs to capture its fair share - and use that demand to accelerate development of its natural resources and enhance regional resiliency.”
Simon Currie, Executive Chair, Kakariki Renewables / Founder & Chief Project Officer, Energy Estate
Energy Estate (EE) is one of four founding partners of Kākāriki Renewables. Through its Energy Estate Digital platform, EE is developing new subsea data cables between California and New Zealand, along with energy and infrastructure precincts in strategically located regions with direct access to renewable energy. The precincts will support the development of data centre campuses and other energy intensive industries.
#1 New Zealand is too small and too remote to be a serious AI data centre hub
Reality: AI does not care where it is trained - it cares about access to power, cost, stability and connectivity.
A critical shift is underway in the AI value chain. While latency matters for consumer facing services, AI model training, batch inference and GPU-as-a-service workloads are increasingly location-agnostic. These workloads can be located wherever electricity, land, cooling and fibre are accessible and competitive.
Boston Consulting Group (BCG) recently published its report Data Centres as Strategic Infrastructure: Unlocking Value for NZ Inc (BCG DCs as Strategic Infra Report). BCG frames data centres as strategic enabling infrastructure: once a country can deliver reliable, scalable electricity, grid access and connectivity, it can “export” digital services the same way it exports physical commodities like dairy, meat and wool.
In other words, remoteness is not disqualifying for AI training and batch workloads; what matters is whether New Zealand can reliably convert its energy advantage into reliable compute at scale - and connect it to global networks.
New Zealand is not “small” in infrastructure terms. It has:
A land area larger than the UK
Industrial scale existing electricity transmission networks
World‑class hydro, wind and geothermal resources
Multiple international subsea cable systems - with the prospect of new international connectivity, including Google’s Honomoana cable, SSCN Tasman Express Southern Cross cable, Datagrid’s Te Waipounamu cable landing at Oreti Beach, and Energy Estate Digital’s CaliKiwi cables with five landing points along the east coast of New Zealand.
#2 New Zealand already has a highly renewable grid, so AI data centres add unacceptable strain
Reality: Data centres are not a threat to the energy transition - they can accelerate it.
New Zealand generates ~90% of its electricity from renewable sources, but electricity represents only ~30% of total energy consumption which takes in account fossil fuel demand from sectors such as transport and industrial and domestic heat.
AI data centres:
Create long dated, consistent high load, bankable demand
Improve the economics of new wind, solar, storage and transmission
Enable co‑investment in generation rather than crowding it out
One of the core points in the BCG DCs as Strategic Infra Report is an “energy abundance” mindset: the fastest way to improve affordability and resilience is to build firmed renewables and the grid ahead of demand, rather than continue to manage scarcity. That framing aligns with Kākāriki Renewables’ “600% Renewables” campaign - the idea that we can build enough renewable electricity to not only decarbonise the grid, but to electrify transport and industrial heat and support new export industries.
Data centres can be a catalyst because they create large, long dated load that helps justify earlier and sustained investment in generation, storage and transmission - reducing the risk that grid and consenting timelines become the binding constraint in the 2030s.
Internationally, many jurisdictions are explicitly using data centre demand to underwrite renewable overbuild and grid resilience. New Zealand has the same opportunity - particularly in regions like Southland with surplus transmission capacity originally built for the Tiwai Point aluminium smelter and concrete plans from Transpower and PowerNet to augment the networks to support electrification of existing industry and attract new energy intensive industries.
#3 AI data centres deliver little local value beyond construction jobs
Reality: The value of developing a data centre industry is not just jobs - it is export infrastructure which can power regional economies. It is true that modern data centres are highly automated. But this misses the point.
AI data centres are export infrastructure, analogous to ports, aluminium smelters or LNG plants - except they export compute instead of meat, metals or molecules. Strategically located and designed AI data centres can:
Attract new sources of local and global capital to New Zealand
Drive electricity transmission, fibre and generation upgrades
Accelerate the development of downstream digital, research and advanced manufacturing ecosystems
The opportunity for New Zealand is capturing the maximum value locally. The power for the data centre will be produced and paid for locally. Regions like Southland have a strong local construction industry which can participate in the data centre build-out and the associated investments in renewable energy.
BCG estimates data centres represent a $70 billion strategic opportunity for New Zealand to 2035 - even without NZ becoming a top-tier global hub. BCG predicts that total new data centre capacity to 2035 (domestic growth plus export demand) could conservatively be ~600MW, translating to incremental electricity demand of up to ~3.5TWh to 2035.
With that framing, the value is not “headcount” - it is the platform effect: once built, data centres pull through investment in generation, transmission and fibre, and enable downstream digital and industrial ecosystems and associated highly skilled jobs.
#4 Cooling, water use and environmental impacts make AI data centres unsuitable for New Zealand
Reality: New Zealand is structurally advantaged for low-impact AI infrastructure.
AI workloads are energy-intensive, but New Zealand’s climate and resource profile can materially reduce their footprint:
Cool ambient temperatures lower cooling demand
Robust water resources across much of the country which can be enhanced through investment in new transfer and treatment facilities
High renewable penetration reduces embodied emissions
Datagrid’s approved Environmental Management Plan explicitly addresses heat rejection, water treatment and discharge controls, reflecting regulatory scrutiny aligned with international best practice.
Compared to existing hyperscale markets in the US, Europe or Singapore - where water scarcity and grid congestion are becoming binding constraints - New Zealand starts from a position of abundance.
The opportunity for New Zealand is to be a world leader in the integration of AI data centres with existing water infrastructure and supporting the upgrades needed to ensure long term water security and affordability for communities and other industries.
The shift away from traditional evaporative cooling (which consumes large volumes of water for heat rejection) in AI data centre design toward closed-loop and liquid cooling architectures is actually reducing consumptive water use i.e. water that evaporates and is lost.
Modern direct liquid cooling (DLC) and immersion cooling systems recirculate coolant in closed loops, dramatically cutting the water that actually leaves the system. For AI-specific workloads (dense GPU clusters), liquid cooling is increasingly the only practical solution and modern implementations are far more water-efficient than legacy air-cooled designs with cooling towers.
For decades, the Marsden Point refinery acted as a major “anchor customer” for Whangārei’s water infrastructure, contributing a large share of rates and water‑use revenue that helped cover fixed costs and keep prices lower for other users.
When the refinery closed in 2022, water demand and revenue fell sharply, but the largely fixed costs of operating, maintaining and financing the network remained, forcing those costs to be spread across a much smaller base of households and businesses. This led to upward pressure on rates and water charges, not due to higher spending, but because the refinery’s cross‑subsidy effect disappeared, demonstrating how critical it is to include the presence and loss of industrial users in regional economic planning.
So rather than being a burden, large industrial users, such as data centres, can stabilise infrastructure pricing and reduce cost of living for a whole region.
#5 AI data centres undermine energy affordability for households and industry
Reality: Poor market design causes price risk – not demand growth.
Electricity price volatility in New Zealand reflects decades of lack lustre demand growth, underinvestment and policy uncertainty, not excess demand.
In 2024 we saw demand destruction and the closure of many longstanding regional businesses – driven in part by the surging electricity prices. The risks of under investment in new generation is exacerbated by the rapid decline in domestic gas supplies and its impact on households and businesses.
There is currently a multi-GW pipeline of wind, solar and battery projects in development across NZ with over 1GW already in construction. There are also viable opportunities to build new large scale geothermal power plants – which was highlighted in the NZ Government’s consultation document released last year - From the Ground Up - A draft strategy to unlock New Zealand’s geothermal potential.
In February this year, the Government entered the party by announcing its plan to use its energy needs to promote new energy generation, encourage innovation, and boost competition in the market. This will be achieved through power purchase agreements with the three central government agencies that consume the most electricity.
The alternative – stagnation – risks higher prices, greater fossil fuel reliance and continued exposure to global energy shocks.
#6 Global tech companies will capture all the value
Reality: Value capture is a policy and partnership choice.
Concerns raised by a variety of academics and commentators following the Makarewa approval have focussed on who controls, benefits from and governs digital infrastructure.
But this is not an argument against data centres - it is an argument for:
Local generation ownership
Māori and regional equity participation
Community benefit frameworks
Sovereign data, compute and skills policies and a forward-looking regulatory framework
New Zealand has successfully navigated similar issues in energy, forestry and infrastructure before. AI data centres are no different – but they require intentional design. The New Zealand Government, EDA’s, data centre developers, and community need to come together quickly and design the public interest principles which best suit New Zealand.
If you are interested in what is going on across the Tasman in Australia (currently #2 market globally for data centre investment), have a look at the recently released Public Interest Principles for Data Centres from a group of NGOs and energy industry bodies.
Let’s compare New Zealand with Ireland. Data centres consume over 20% of Ireland’s electricity and new data centres have faced opposition and a moratorium. Many factors have contributed to Ireland becoming Europe’s digital hub – such as location, tax rates, investment attraction framework and relative speed of planning approvals. New Zealand doesn’t share all these factors.
But let’s stand back and look at the big picture. In 1995 the GDP per capita in each of Ireland and New Zealand was hovering around USD20,000. These days Ireland is over USD100,000 and New Zealand has not broken through USD50,000.
New Zealand’s natural resources need to be harnessed for more than exporting dairy and other agricultural products and attracting tourists for the perfect Instagram post. Data centres should be viewed as the start and not the finish line – and as the enabling infrastructure for opportunities in an electrified future.
A Step Towards a More Prosperous Future
The approval of Datagrid’s Makarewa AI data centre is not an outlier - it is a step into towards a more prosperous future.
BCG’s “strategic infrastructure” framing is useful here: data centres sit alongside transmission, generation and subsea cables as foundational capacity that supports and attracts other industries such as manufacturing, advanced food processing and campuses.
Put simply, when New Zealand builds data centres, it is not just new buildings housing computers - it is expanding the country’s ability to monetise renewable energy as a globally traded digital product and accelerating the build-out of new generation capacity and enhancement of one of the world’s cleanest and most resilient national grids.
Global AI infrastructure is expanding faster than traditional markets can supply power, land and permits. New Zealand offers something increasingly rare. BCG estimates that every $1 invested in renewable generation to support data centres can unlock roughly $13–18 of economic impact over ten years, through the combined effects of generation investment, data centre capex, operations and supply chain activity.
New Zealand offers the AI industry a range of attractive features such as:
Clean, firm electricity
Cool climate
Political stability and geographic diversification
Clear pathways to gaining and maintaining social licence for the industry
As with renewable energy, the greatest risk is not moving too fast — it is failing to recognise the scale of the opportunity in time and falling further behind.