To suggest we can build an economy by exporting electrons is fundamentally screwed
A nation at the edge of the Pacific is weighing whether its cool climate, clean energy, and political calm can be converted into a new kind of wealth — one measured not in goods shipped but in computations performed. New Zealand's crown investment agency is courting $25 to $30 billion in offshore capital for data centres, hoping to position the country as an artificial intelligence hub for Asia while the world's existing compute capacity strains under demand. The promise is jobs, export earnings, and accelerated AI adoption; the question is whether the benefits will stay onshore, or whether the country will absorb the costs — in power prices, water, and land — while the profits flow elsewhere.
- Global demand for data centre capacity has saturated markets in the US, Malaysia, and Ireland, and cities from Seattle to Denver have imposed moratoriums — creating a window New Zealand is racing to fill before the moment passes.
- Invest NZ CEO Robert Wall is pitching the country's mild 13.5°C average, high renewable electricity share, and single national grid as rare advantages that make large-scale hyperscale development both feasible and defensible.
- Critics warn the model is structurally broken: hyperscalers shift profits offshore, contribute minimally to local tax revenue, and risk driving up electricity prices for the manufacturers and households already struggling to stay viable.
- On-the-ground evidence is sobering — Amazon abandoned its Auckland site after purchasing it, Microsoft's second planned facility never progressed, and Datagrid's $5 billion Southland project remains unfunded despite winning resource consents.
- The arithmetic Wall offers is stark and simple: if the full investment materialises, expect one job per megawatt built — a figure that must now convince a public watching the same backlash unfold in Britain and across the United States.
Robert Wall returned to New Zealand in February after thirty years abroad, stepping into the chief executive role at Invest NZ with a mandate as large as any the crown agency has attempted: attract between $25 and $30 billion in foreign capital for data centres and the infrastructure surrounding them. His background — civil engineering, investment banking, and most recently sustainable private infrastructure at Lazard in London — shaped a pitch built around geography and stability. New Zealand, he argues, can become an artificial intelligence compute hub, selling excess processing power to Asia at latencies that physics permits, while accelerating domestic AI adoption and building what he calls a weightless export sector.
The case rests on a cluster of genuine advantages. A mild average temperature of 13.5 degrees Celsius reduces cooling costs. A high share of renewable electricity and a single national grid operator make orderly, large-scale development possible in ways that fragmented markets cannot match. Wall also points to existing policy levers — an R&D tax credit, accelerated depreciation through the Investment Boost, and fast-track consenting — as signals of regulatory seriousness. The global timing matters too: capacity in the United States, Malaysia, and Ireland has been absorbed, and community opposition has forced moratoriums in Seattle, Baltimore, Denver, and dozens of American counties, pushing hyperscalers to look elsewhere.
Yet the local record is uneven. After Jacinda Ardern's 2020 visit to Seattle, both Amazon and Microsoft received approval to build hyperscale facilities in Auckland's northwest. Microsoft completed a roughly $1 billion facility at Westgate in late 2024, but its second planned site never advanced and the consent lapsed. Amazon bought land, then abandoned it, opting instead to co-locate with existing operators. Datagrid's $5 billion Southland project has resource consents but no funding and is still awaiting grid approval for its first 280-megawatt stage — with ambitions to eventually reach 1000 megawatts, equivalent to 13 percent of the country's current generating capacity.
The sceptics are pointed. Don Christie of Wellington-based Catalyst argues that hyperscalers are structurally designed to minimise local tax contributions, shifting revenue and profit offshore while leaving communities to absorb the costs. He draws a sharp distinction between data centres built for New Zealand's own needs — the model pursued by local operators like Datacom and T4 Group — and facilities built to export compute to offshore clients. His deeper concern is electricity: New Zealand's manufacturing sector is already closing because power prices are too high, and adding enormous new loads to serve foreign customers risks compounding that damage. Wall's response is that construction activity alone has economic value, and that renewable power agreements tied to new projects will expand generating capacity rather than simply divert existing supply. Whether that argument holds — and whether New Zealand avoids the community backlash now reshaping the data centre landscape across the English-speaking world — is the question the next few years will answer.
Robert Wall arrived back in New Zealand in February after three decades abroad, taking the helm of Invest NZ just as the crown agency was preparing to chase one of the country's largest foreign investment targets: between $25 and $30 billion in offshore capital for data centres and the infrastructure that feeds them. Wall's résumé reads like a blueprint for the job—eight years in civil engineering, then investment banking, most recently as managing director of sustainable private infrastructure at Lazard Asset Management in London. He came home with a pitch: New Zealand could become a hub for artificial intelligence compute, selling excess processing power to Asia at speeds that geography permits, while accelerating the country's own AI adoption and creating what he calls a weightless export sector.
The ambition is not small. Microsoft's proposed West Auckland facility would bring 300 temporary construction jobs and 50 permanent positions once operational, according to its Overseas Investment Office application. Datagrid's $5 billion Southland project promises 75 direct jobs, with another 72 supported through suppliers and servicing—modest compared to the thousand workers at Rio Tinto's nearby aluminium smelter, but Wall argues the construction phase alone matters. "The US would be in recession right now if it wasn't for data centre construction," he told media before announcing the investment push. He pitched New Zealand's advantages methodically: a mild climate averaging 13.5 degrees Celsius, a high percentage of renewable electricity, wholesale power prices trending downward, peace, stability, and a single national grid operator that makes orderly development possible. He also highlighted the R&D tax break, the Investment Boost for accelerated depreciation, and fast-track consenting measures already in place.
Yet the global context is shifting. Capacity in the United States, Malaysia, and Ireland has been absorbed. Several American cities have turned hostile. Seattle imposed a year-long moratorium after fears that five proposed hyperscale facilities would consume a third of the city's power. Baltimore, Reno, Minneapolis, Tulsa, and Denver followed suit. Counties in Georgia, Michigan, North Carolina, and Ohio did the same. The opposition stems partly from legitimate concerns: data centres consume vast amounts of electricity and water, and in less regulated markets, operators have driven up consumer power bills and used evaporation cooling that depletes local water supplies. New Zealand's stricter environmental standards and cooler climate—which makes water recirculation feasible—offer advantages Wall believes can be defended locally.
But scepticism runs deep here too. Don Christie, managing director of Wellington-based Catalyst and co-founder of the now-defunct advocacy group NZRise, argues the model is fundamentally flawed. The hyperscalers shift revenue and profit offshore, minimising local tax contributions. "If we build data centres for New Zealand, on a New Zealand scale, which is what T4 Group and Datacom and others have been doing, that's a completely different thing," Christie said. He contends the country should invest instead in homegrown AI startups that could rival Rocket Lab, rather than exporting electrons to offshore clients powered by fossil fuels. He also points to a more immediate problem: New Zealand's manufacturing industries are closing because power prices are too high. "To suggest that we can build an economy by exporting electrons is just fundamentally screwed," he said.
The track record offers mixed signals. In 2020, after then Prime Minister Jacinda Ardern visited Seattle, Amazon and Microsoft both received OIO approval to build hyperscale data centres in Auckland's northwest. Microsoft spent around $1 billion on a facility at Westgate that opened in December 2024, but its twin at Whenuapai never progressed beyond initial planning and the company lost its time-limited consent. Amazon abandoned its giant Westgate site months after buying its final parcel of land, shifting instead to co-locating with existing operators. DCI and CDC Data Centres did build hyperscale facilities in the first half of this decade, but the momentum has stalled. Datagrid's Southland project has won resource consents but remains unfunded and awaiting Transpower approval for its 280-megawatt first stage. Spark's data centre north of Auckland, now owned by TenPeaks after the telco sold 70 percent of its data centre business to an Australian private equity firm, has been on the drawing board for years.
Wall did not offer an immediate explanation for why Amazon and Microsoft had stalled their own builds. Mercury has signed an option to supply Datagrid with 140 megawatts over 15 years, and the national grid operator says there is enough capacity in the pipeline to facilitate the $5 billion facility, but Datagrid's owners ultimately want to scale to 1000 megawatts—equivalent to 13 percent of New Zealand's current generating capacity. That ambition depends on new renewable power plants coming online and a companion international cable to export processing power. Wall's final metric was simple: if the $25 to $30 billion materialises, Invest NZ expects one job per megawatt of capacity built. Whether that calculus persuades a sceptical public, or whether community opposition rises as it has in the United States and Britain, remains to be seen.
Citações Notáveis
The US would be in recession right now if it wasn't for data centre construction— Robert Wall, Invest NZ CEO
If we build data centres for New Zealand, on a New Zealand scale, that's a completely different thing. We could be world-leading in AI if we invest in start-up businesses instead— Don Christie, Catalyst managing director
A Conversa do Hearth Outra perspectiva sobre a história
Why does Wall think New Zealand can succeed where so many American cities are now saying no?
He points to three things: our cooler climate makes water recirculation work without depleting supplies, we have stricter environmental standards than most countries, and we have a single grid operator instead of fragmented utilities. That means a more orderly build-out. But he also admits opposition will likely rise as construction intensifies.
The power price argument seems to cut both ways. Wall says data centres will fund new renewable capacity, but critics say that's not what happened in Europe.
Right. The theory is that a data centre signs a 15-year power purchase agreement with a generator, who then builds new capacity. Both take two to three years. But Christie argues the hyperscalers have already captured our renewable capacity on the cheap, and we're left topping up via coal-fired Huntly. There are examples both ways—Amazon and Microsoft did sign deals that funded new wind and geothermal—but the scepticism is rooted in who captures the value.
What happened to the Amazon and Microsoft projects that already got approval?
That's the puzzle nobody quite answers. Microsoft built one facility at Westgate for about a billion dollars, which opened last year. But its twin at Whenuapai never happened and the consent expired. Amazon bought a massive site in Westgate, did some earthworks, then abandoned it and shifted to co-locating with existing operators instead. Wall didn't offer a theory.
So what's actually being built right now?
CDC Data Centres built hyperscale facilities in Hobsonville and Silverdale in the first half of this decade. But the newer projects—Datagrid in Southland, TenPeaks' facility north of Auckland—have been stuck on the drawing board for years. Datagrid has consents but no funding confirmed and no Transpower approval yet. It's waiting for new renewable plants to come online.
What does Christie think should happen instead?
He thinks New Zealand should build data centres at a New Zealand scale, like T4 Group and Datacom do, and invest in homegrown AI startups that could be the equivalent of Rocket Lab. He sees exporting electrons to offshore clients as a dead end—especially when our own manufacturing industries are closing because power is too expensive.
Is Wall's one-job-per-megawatt estimate realistic?
It's a projection, not a promise. It assumes the $25 to $30 billion actually lands and gets built. Given that Amazon and Microsoft have already stalled their own projects, and Datagrid has been waiting for years, there's reason to wonder if the global appetite is as strong as Wall believes.