New Zealand is racing to attract AI infrastructure investment by selling its renewable electricity, cool climate, and political stability. But an analysis from the University of Auckland published this week argues the strategy rests on borrowed lessons from countries that succeeded through means NZ isn’t replicating — and risks turning the country into infrastructure for someone else’s AI economy.
The critique, published on 20 August, takes aim at the government’s “concierge” approach to foreign investment. Prime Minister Christopher Luxon’s Invest New Zealand agency, modelled on Ireland and Singapore, is designed to “roll out the welcome mat” for foreign investors. The analysis questions whether that mat leads anywhere worth going.
The Ireland and Singapore problem
The argument is straightforward. Singapore didn’t become a hub by opening its doors. It became one through state-led investment and equity stakes across key industries — banking, aviation, ports, telecommunications — backed by sovereign wealth funds like Temasek and GIC. Changi Airport and the Jurong Industrial Estate weren’t accidents. They were built deliberately as foundations.
Ireland attracted Google, Microsoft, and Amazon through low taxes and EU membership. But Big Tech built real European operations in Dublin’s Silicon Docks — not just profit-booking shells. The NZ analysis argues the current government is borrowing the geography and investor facilitation parts of those models without the public investment and economic coordination that actually made them work.
What NZ’s data centre experience already shows
The paper points to New Zealand’s recent experience with hyperscale data centres from AWS and Microsoft as a warning. Those investments came with promises of supercharged local tech ecosystems. What actually happened, according to the analysis: Big Tech reduced its tax burden by routing revenue through Irish parent companies and using inter-company service fees to shift hundreds of millions offshore.
The promised economic multipliers — jobs, exportable digital industries, global competitiveness — remain largely hypothetical. Smaller local cloud providers struggle to compete with Big Tech on price and influence, and are increasingly pushed toward the physical side: securing land, building facilities, hosting the digital services of the same multinationals they used to compete with.
This connects to findings from our earlier reporting on NZ’s data centre build-out and who actually benefits, and the broader question of whether NZ should build its own sovereign AI stack rather than host someone else’s.
Datagrid as a case study
The analysis singles out Datagrid’s proposed AI data centre near Invercargill — a project we’ve covered in detail — as where these concerns converge. Datagrid is Singaporean-owned. Its tax practices are unknown. For a multi-billion-dollar investment, it projects a relatively low number of long-term employees. And its customers are likely to be offshore AI and cloud providers.
That positioning matters. If the customers are overseas, the high-value work — the AI services running on top of the compute — happens elsewhere. NZ provides the land, the electricity, and the cooling. The strategic decisions and the higher-value digital services stay offshore.
The phrase the University of Auckland analysis uses is pointed: without stronger public stakes or local capability requirements, the “concierge” strategy risks turning New Zealand into “a well-powered refrigerator for the global AI economy.”
A different direction across the Tasman
The analysis highlights a recent speech by Australian Prime Minister Anthony Albanese as a counterpoint. Albanese framed AI infrastructure as an industrial strategy issue — linking labour, copyright, national security, energy, water, and regional development — rather than simply an investment attraction question.
The Albanese speech explicitly raised the politics of ownership and control, arguing against “subcontracting our sovereignty and security to the control of foreign monopolies” and pointing to Australia’s national broadband network as a model of public ownership over foundational digital infrastructure.
Whether Australia follows through is another question. But the framing is materially different from NZ’s approach, which the analysis describes as pulling in two directions: a concierge service for foreign investment on one hand, and MBIE’s AI strategy prioritising adoption of existing models over domestic development on the other. Taken together, NZ is positioned as a customer and a host — but not as an owner or producer.
What the analysis doesn’t address
The critique focuses on the economic case. It doesn’t deeply engage with the counter-argument that access to Big Tech AI services is itself enabling for domestic firms — cheaper compute, sophisticated tools, no need to build from scratch. The analysis acknowledges this but argues access isn’t the same as developing domestic capability. When foreign firms control the platforms and standards, they also shape the terms on which local companies compete.
It also doesn’t address the environmental case, which others have made separately — that NZ’s renewable grid makes it a relatively clean place to run AI compute, regardless of who owns the infrastructure. That argument has its own complications, as we’ve explored in our coverage of Datagrid’s carbon arbitrage framing.
🔍 THE BOTTOM LINE
The University of Auckland analysis raises a question the government’s investment prospectus doesn’t answer: what does NZ actually own at the end of this? If the strategy is clearing a path for foreign capital to build infrastructure that serves overseas customers, the economic spillover may be thinner than the renderings suggest. The Ireland and Singapore models worked because those governments invested alongside foreign capital, not just facilitated it. NZ’s current approach borrows the welcome mat without the foundation.
❓ FAQ
What is NZ’s “AI hub” strategy?
The government, through Invest New Zealand, is promoting NZ as a destination for AI infrastructure investment — data centres, compute facilities — by highlighting the country’s renewable electricity, cool climate, and political stability. The pitch targets a potential $25–35 billion “AI hub” market.
Why does the University of Auckland critique matter?
It’s one of the few detailed academic analyses of the economic case for NZ’s AI infrastructure strategy, published by the country’s leading university. It directly challenges the government’s framing and offers an alternative vision based on public investment and domestic capability.
What did Australia’s PM say about AI infrastructure?
Anthony Albanese delivered a speech framing AI infrastructure as an industrial strategy issue, linking it to labour, copyright, national security, energy, and regional development. He argued against subcontracting sovereignty to foreign monopolies and pointed to the national broadband network as a model of public ownership.
Does this mean NZ shouldn’t attract AI investment?
The analysis doesn’t argue against foreign investment. It argues that foreign investment alone — without public stakes, local capability requirements, or mechanisms to retain value — won’t make NZ an AI hub. The question is what NZ gets to own and build, not whether it should participate.
