The lazy consensus of the political class is that we can regulate our way out of a physics problem. When federal leaders and climate advocates hyperventilate over a projected seven-fold or ten-fold spike in data centre power consumption, they treat computing demand like an adjustable dial on a kitchen stove. Just twist the knob down, mandate a few local green energy offsets, and pretend the grid will remain peaceful.
It is a comforting delusion. And it is entirely wrong.
The panic over server racks gobbling up regional power supplies rests on a fundamental misunderstanding of how digital infrastructure scales. Governments think they are dealing with a standard industrial zoning issue. They assume that by forcing developers to jump through localized planning hoops, build away from schools, or pledge token renewable contributions, they can protect ordinary power bills.
I have watched companies burn millions trying to negotiate compromise energy frameworks in constrained markets. The dirty secret nobody in Canberra or state parliaments wants to admit is that compute demand does not bargain. If an artificial intelligence model requires gigawatts of continuous, uninterruptible power to train or query, it will go wherever that power exists. If you build bureaucratic walls around domestic markets, the capital simply flows offshore, leaving behind a hollowed-out tech sector and a stagnant economy.
The Flawed Premise of Localized Energy Capriciousness
Let us dismantle the core argument driving the current legislative panic. Critics point to Australian Energy Market Operator forecasts showing consumption scaling dramatically over the coming decade and scream for immediate moratoriums. They argue that every megawatt consumed by a server rack is a megawatt stolen from a suburban household or a local hospital.
This is a zero-sum fallacy built on economic illiteracy.
Data centres are not traditional factories belching smoke while producing widgets for local consumption. They are the foundational settlement layer of the modern global economy. Treating them like nuisance factories is like trying to regulate the internet by telling telecom providers how many copper wires they are allowed to string past a paddock.
Imagine a scenario where a state government successfully blocks major computing infrastructure through punitive zoning laws and rigid local mandates. What happens next? The capital does not stay put. It migrates to jurisdictions with fewer scruples and more flexible energy grids. The emissions are not prevented; they are merely exported. Meanwhile, the local population is left stranded on an isolated, expensive grid that lacks the industrial tax base required to upgrade aging transmission lines.
The real question is not how to starve these facilities of electricity. The question is how to use the immense, concentrated balance sheets of tech conglomerates to force a total overhaul of national power generation capacity.
The Hard Truth About Grid Economics
The anti-infrastructure crowd loves to claim that server hubs will inflate household electricity prices. They assume these massive energy consumers will free-ride on existing public investments.
In reality, massive compute operations are the only entities with enough financial leverage to underwrite utility-scale baseload solutions. They do not want intermittent, unreliable trickle power. They need dedicated generation. When tech giants build or contract power, they look for massive scale. If policy forces them to fund entirely new generation assets from scratch, those assets remain on the grid long after the initial server racks are obsolete.
The danger of current regulatory overreach is that it treats every gigawatt of demand as an enemy rather than an investment trigger. If you want a decarbonized grid, you need capital at a scale that governments refuse to appropriate directly. Private infrastructure deployment funded by corporate balance sheets is the engine that actually builds transmission lines and storage arrays.
The Uncomfortable Downside
My argument has a dark side, and pretending otherwise is dishonest.
Unfettered expansion puts immense localized strain on water resources and high-voltage distribution corridors. Cooling massive computing clusters requires extreme engineering. Communities hosting these facilities often experience acute noise pollution, strained local housing markets due to construction booms, and immediate transmission bottlenecks before new generation assets come online.
There is no painless transition. Allowing digital infrastructure to expand requires accepting friction in host communities. Pretending we can craft a legislative silver bullet that eliminates all environmental footprint while retaining cutting-edge economic growth is a political fairy tale.
We are forced to choose between two realities. We can accept the messy, capital-intensive reality of rapid infrastructure buildout, or we can choose managed decline behind a wall of protective red tape.
Stop trying to legislate away the laws of thermodynamics. Align the incentives, force massive private investment into generation, and get out of the way before the rest of the world leaves your grid in the dark.
