Tasmania’s Power Puzzle: Greens Push Inquiry Over Foreign AI Data Centre Boom
Tasmanian Greens have forced an urgent parliamentary inquiry into how large, foreign-owned AI data centres are being approved — with no meaningful regulatory scrutiny and no guarantee the state benefits. The core question is stark: should Tasmania hand its finite, renewable hydro capacity to overseas AI operators running NVIDIA GPU clusters, while the industries that actually employ Tasmanians wait years for grid upgrades?
🔍 THE BOTTOM LINE
This isn’t a debate about AI. It’s a debate about who gets the electrons. Singaporean operator Firmus has secured 104MW in Launceston — roughly the entire power draw of the Boyer paper mill, which employs 300 people. Boyer can’t get the upgrade it needs to modernise. A data centre drawing the same energy will create 6 to 8 permanent jobs. The Greens are asking the question every renewable-rich jurisdiction should be asking: if your clean energy is a finite public resource, why are you giving it away to foreign compute with almost nothing coming back?
The Scale of the Energy Demand Conflict
The numbers don’t sit comfortably together. In 2024-25 Tasmania was a net electricity importer — pulling in 1983 GWh from Victoria (much of it brown-coal-fired) while exporting only 433 GWh back. The state that markets itself as 100% renewable is, in practice, leaning on mainland coal to cover shortfalls. Against that backdrop, handing 104MW of continuous load to a single foreign data centre operator is not a marginal decision.
Dr Paris Buttfield-Addison lays out the asymmetry plainly: the grid was built to serve Tasmanian industry and households, not to become a cheap clean-power substrate for global AI inference. When local industrial users face connection bottlenecks measured in years, but foreign data centre operators appear to move from proposal to approval at pace, the impression — fair or not — is that there are two queues, and the one with foreign capital in it moves faster.
This is the same power bottleneck Aschenbrenner flagged: AI’s hunger for electricity is now the binding constraint, and it is colliding with grids that were never sized for it. Tasmania is just a particularly sharp instance of a global pattern.
Jobs vs. Gigawatts: The Employment Argument
The employment maths is brutal. Boyer paper mill: ~300 jobs. Bell Bay aluminium: hundreds more. A large AI data centre: typically 6-8 permanent positions, plus a construction crew that leaves when the building goes up. That is the exchange on offer — replace hundreds of existing, unionised, regional jobs with a handful of facilities jobs tending racks of GPUs owned overseas.
Greens leader Dr Rosalie Woodruff has pressed this point repeatedly in Parliament: why are foreign operators getting large power allocations signed off while the state’s own industrial base can’t get the grid capacity to expand or even sustain operations? It’s not an anti-technology stance. It’s a question about who the grid is for.
The exchange rate is worse than it looks once you account for what economists call embedded benefit. A paper mill buys local timber, pays local rates, trains local apprentices, and anchors a town. A data centre imports hardware, runs it lights-out, and repatriates revenue offshore. The 6-8 jobs are real, but they are thin — and they don’t regenerate a regional economy the way a mill does. For more on this pattern across the Tasman, see our coverage of the Australian data centre backlash.
Regulatory Oversight and Local Impact
The inquiry is fundamentally about process. Tasmania’s current approval pathway for large loads does not appear to require a rigorous test of whether a given connection serves the state’s economic interest. There is no clear gate that asks: does this allocation create more value, in jobs and resilience, than the alternative use of that megawatt? Without that gate, decisions default to whoever brings capital and asks first.
The Five Eyes cybersecurity warning this week — which flagged frontier AI models as a near-term national-security risk — sharpens the governance question rather than softening it. If foreign-operated AI infrastructure is now also a security consideration, then rubber-stamping large foreign load connections without an interest test looks not merely economically careless but strategically complacent. As The Guardian reported, the joint Five Eyes statement explicitly warned that frontier AI “accelerates the speed, scale, and sophistication of cyber threats.” Hosting that workload — under foreign ownership, on a state grid — without a public-interest test is a choice, not an inevitability.
NZ Angle
New Zealand faces the identical question, only sharper. Tasmania’s grid is small; New Zealand’s is smaller still — roughly 45 TWh annual generation versus Tasmania’s ~12 TWh, but serving a population five times larger and with far less surplus headroom per capita. When a single 100MW+ AI load lands in NZ, it doesn’t nibble at the margin. It materially reshapes what else can connect, and at what price.
The parallels are uncomfortable. NZ’s renewable mix — hydro in the South Island, geothermal in the Bay of Plenty, wind scattered between — is exactly what foreign AI operators want: clean, firm-ish power they can bolt onto ESG disclosures. The same foreign-capital-asks-first dynamic is visible here. Local industrial users, from dairy processing to data-hosting NZ firms to new green-hydrogen projects, queue for connection studies while overseas operators negotiate directly for tens of megawatts of firm renewable supply. As we noted in NZ has no plan for data centres, there is no national-interest framework in this country for deciding who gets priority on a constrained grid. The default is whoever pays.
The trade-off is starker here precisely because the grid is smaller. Tasmania can, at the margin, import from Victoria’s coal fleet when hydro runs short. New Zealand has no mainland to lean on — the HVDC link balances North and South Island, it doesn’t add net capacity. Every megawatt handed to a foreign AI operator is a megawatt a local industrial user cannot take. There is no plan for adjudicating that choice, and until there is, decisions will continue to be made by inertia and by chequebook.
❓ FAQ
Q1: How much power does the Firmus data centre in Launceston require, and what’s the comparison? A: Firmus is contracted for approximately 104MW — roughly equal to the entire operational draw of the Boyer paper mill, which employs around 300 people. The data centre is expected to create 6-8 permanent jobs.
Q2: Isn’t Tasmania fully renewable? Why is there a supply conflict? A: In 2024-25 Tasmania was a net importer: 1983 GWh in from Victoria (largely brown-coal and gas generation), only 433 GWh exported back. The “100% renewable” headline obscures real shortfalls, especially in dry years when hydro output drops.
Q3: How many jobs does a large AI data centre actually create? A: Typically 6-8 permanent operational roles per site, plus a temporary construction workforce. By contrast, the Boyer mill and Bell Bay aluminium smelter each anchor hundreds of direct, unionised, regional jobs plus substantial indirect employment.
Q4: What exactly is the Greens’ inquiry demanding? A: A parliamentary review of the approval process for large grid connections — specifically whether foreign-owned AI data centres are being signed off without a public-interest test weighing job creation, local supply-chain benefit, and grid resilience against the alternative industrial uses of that capacity.
Q5: Does NZ face the same problem? A: Yes, and more acutely. NZ’s grid is smaller than Tasmania’s with no mainland fallback, so a single 100MW+ foreign AI load materially reshapes who else can connect. There is currently no national-interest framework in NZ for adjudicating priority on a constrained grid.
🔍 THE BOTTOM LINE
Tasmania is at an inflection point where global AI compute demand is colliding with finite local resources — and the deal on the table is a bad one. A 104MW foreign data centre buys you 6-8 jobs and repatriated revenue; the same megawatts could secure 300 existing industrial jobs and a regional economy that actually regenerates. The Greens’ inquiry is not political theatre; it is a necessary intervention to force a public-interest test onto decisions that have, until now, been made by whoever brings capital and asks first. New Zealand should be watching closely — the same calculus is coming for our grid, our geothermal and our hydro, and we have even less margin to give away.
