Greens want to stop the data-centre rollout.
AEMO's pipeline of data centre projects doubled in a year. The Greens want to pause all of it — but the costs of that call are hiding in plain sight.
Greens want to stop the data-centre rollout.
The Victorian Greens have responded to updated Australian Energy Market Operator forecasts by demanding an immediate moratorium on new large-scale data centres, citing a projected jump in national electricity consumption from these facilities to 34 terawatt hours by 2036, equivalent to roughly 13 per cent of the grid. This surge, they argue, undermines efforts to replace coal, electrify the economy and lower power bills. Their call for a pause until proper rules govern energy, water and planning impacts comes with a parallel push for full release of any modelling on those effects.
The pipeline more than doubled in a year — that is not a fad
The numbers are arresting. AEMO's latest projections show the data-centre connection pipeline has more than doubled in a year, from 97 projects to 225. That scale reflects genuine global appetite for artificial intelligence, cloud storage and digital services, not some passing fad. In Australia this boom has helped drive business investment higher at a time when other sectors have been softer, as noted in earlier analysis of why capital spending has held up despite broader economic headwinds.
Data centres are footloose enough to chase cheap, reliable power — blocking them domestically simply displaces the activity to other jurisdictions or forces it underground in less efficient forms.
A moratorium removes the price signal that drives solutions
Yet the Greens' remedy, halting approvals while "proper rules" are devised, treats the demand itself as the problem. This inverts the incentive structure. Data centres are footloose enough to chase cheap, reliable power. They can co-locate with renewable projects, invest in their own storage, or shift flexible loads to times of surplus generation. Blocking them domestically simply displaces the activity to other jurisdictions or forces it underground in less efficient forms. Markets solve matching problems faster than central planning when price signals are allowed to work. A moratorium removes the signal.
This is not theoretical. Previous coverage of the federal government's insistence that data centres avoid gas entirely showed how such constraints strand investment in regions like the Northern Territory, where renewable resources exist but cannot yet deliver 24-hour firmness at competitive cost. The same tension appears in the Senate inquiry now examining how AI-driven power needs intersect with climate targets. Each time the response has been to tighten the regulatory noose rather than expand supply options. The result is predictable: less investment, slower innovation diffusion, and higher costs borne by the rest of the economy.
The transparency demand is right — but it must apply to renewables too
The Greens are correct, however, to insist on seeing the modelling. Any forecast of 13 per cent of national electricity use deserves scrutiny of assumptions about uptake rates, efficiency improvements, and locational flexibility. Without it, policy drifts on anecdote and press release. But intellectual consistency requires applying the same lens to the renewable transition itself. Renewables have delivered a merit-order effect that suppresses wholesale prices when the wind blows and sun shines. That is real. What the spot price does not reveal are the parallel system costs: $6 billion in new transmission lines to connect remote solar and wind farms to population centres, backup gas plants and batteries that sit idle much of the time, synchronous condensers to replace the inertia once provided by spinning coal turbines, and network upgrades recovered through fixed charges on every bill.
These expenses do not disappear because they are socialised across the grid rather than charged directly to data-centre operators. Germany's experience offers a cautionary parallel. After decades of aggressive renewable deployment, household electricity prices remain among the highest in Europe, driven by levies to cover the infrastructure and intermittency costs that the wholesale market price obscures. The lesson is not that renewables are worthless, but that pretending their system-wide price tag is zero distorts choices elsewhere, including how much computing capacity the economy can afford.
A blanket pause forecloses the market experiments that would actually solve the problem
Victoria's Labor government has indeed rolled out the welcome mat for data centres, seeing jobs and prestige in an industry that employs few people directly but enables many more downstream. That enthusiasm needs tempering by honest accounting. Yet a blanket pause is the opposite of tempering. It freezes the market's ability to experiment with solutions: hyperscale operators funding their own solar farms with battery buffers, district cooling systems that recycle waste heat, or demand-response contracts that treat computing load as a virtual power plant. These innovations emerge under competitive pressure, not during a regulatory time-out.
The deeper risk is that data centres become the scapegoat for a transition whose engineering realities have proved more stubborn than political rhetoric allowed. Electrification of homes, vehicles and industry was always going to lift total demand. Artificial intelligence simply accelerates one slice of that curve. Pretending the curve can be flattened by vetoing new facilities is wishful accounting. It does not reduce the underlying need for more generation, firmer supply and smarter grid management. It merely ensures that Australians capture less of the economic upside while still paying the system costs.
Transparency should be the minimum standard. Release the data-centre modelling. Release the full cost curves for transmission, storage and frequency control. Let markets and voters see the trade-offs without the gloss. Then let investment flow where the numbers make sense. Halting progress on data centres might feel like decisive climate action. In practice it is an admission that the energy system cannot adapt without rationing the future. That is a worse outcome for ordinary households facing higher bills and slower productivity growth than any forecast spike in server-farm demand.
Frequently Asked Questions
How much electricity will Australian data centres use by 2036?
AEMO projects data centres will consume 34 terawatt hours per year by 2036, equivalent to roughly 13 per cent of national grid demand. The connection pipeline for new facilities more than doubled in a single year, from 97 projects to 225.
Why would a moratorium on data centres push investment offshore?
Data centres are footloose: operators choose locations based on power cost, reliability and regulatory conditions. Halting approvals in Australia does not reduce global demand for computing capacity — it redirects that investment to jurisdictions with fewer restrictions, taking the economic benefits with it.
Are data centres really responsible for rising electricity costs in Australia?
Data centres add to total demand, but the primary drivers of household bill increases are network charges that recover the cost of new transmission lines, firming capacity and grid-stabilisation equipment — costs that are socialised across all consumers rather than charged directly to large users. Blaming data centres obscures that structural problem.
What hidden costs does the renewable energy transition add to power bills?
Beyond the wholesale price, the transition requires new long-distance transmission lines, backup gas and battery capacity that sits idle much of the time, and synchronous condensers to replace grid inertia. These system costs are recovered through fixed network charges on every bill, not through the spot price that politicians typically cite.
What is the alternative to a data centre moratorium?
Market-based approaches — allowing operators to co-locate with renewable projects, enter demand-response contracts, or fund their own storage — can match supply with demand faster than a regulatory pause. Full public disclosure of both data-centre energy modelling and renewable system costs would let markets and voters assess trade-offs honestly.