Australia is charging headlong into a new industrial revolution, or so the hype goes. Data centres, the humming cathedrals of the AI age, are sprouting across the country as governments and investors chase the next big thing in the digital economy. Billions are being promised, land is being rezoned, and politicians are falling over themselves to brand the nation a future AI powerhouse. Yet beneath the glossy renderings and economic forecasts lies a stubborn, inconvenient truth: these facilities are voracious energy consumers, and Australia's path to powering them reveals the hollowness of the anti-coal, renewables-only dogma that has dominated policy for years. The data centre boom isn't just exposing the limits of our current energy strategy, it is poised to bury the ideological mania that has left the grid fragile, expensive, and ill-prepared for the future.
The numbers are eye-watering. A single large hyperscale data centre can consume as much electricity as a small city. With AI training and inference demands exploding, projections for Australia's data centre capacity point to multi-gigawatt requirements in the coming decade. That's not a rounding error on the national grid; it's a structural shift that will test every assumption about how we generate, transmit, and price power. Renewables advocates have long promised that wind and solar, backed by batteries, could shoulder the load of a modern economy. The data centre surge makes that promise look increasingly like wishful thinking. Intermittent sources struggle with the always-on, high-density demands of these facilities. Batteries help smooth short gaps but cannot economically replace the baseload reliability needed for 24/7 operations at this scale.
Enter coal, the fuel that dare not speak its name in polite policy circles. Australia sits on some of the world's most abundant and high-quality coal reserves. For decades, it provided cheap, reliable power that underpinned manufacturing, mining, and households. The rush to decarbonise, driven by climate targets and activist pressure, has seen coal plants retired faster than replacements could be built. Gas was supposed to bridge the gap, but domestic prices have soared and exports compete for supply. Renewables have scaled impressively in raw capacity, yet they come with well-documented challenges: transmission bottlenecks, curtailment, and the need for massive overbuild plus storage to achieve anything resembling firm power. The data centre operators, many of them global tech giants with demanding uptime requirements and corporate ESG targets, are quietly discovering that rhetoric and reality diverge sharply when the servers start spinning.
This is where the delusion meets the mirror. Pro-renewables policymakers have bet the grid on a rapid transition that assumed demand growth would be modest and flexible. Data centres blow that assumption apart. Their power hunger is relentless and location-sensitive. They need firm, dispatchable generation that can ramp and run without the variability that plagues solar and wind. Extending the life of existing coal plants, or even building high-efficiency new ones with carbon capture where viable, suddenly looks less like environmental heresy and more like pragmatic engineering. The alternative, relying on yet more intermittent capacity plus enormous battery banks and pumped hydro, comes with eye-watering costs that will ultimately be passed to consumers and industry. Australia risks pricing itself out of the very AI/data centre race it wants to win.
The irony is thick. The same voices that cheered the closure of reliable coal generators and demonised "fossil fuels" are now confronted with the energy appetite of the technologies they celebrate as the future. You cannot run a digital economy on virtue signals and weather-dependent generation alone. Data centres demand predictability, density, and scale, attributes traditional thermal power provides far more cost-effectively than today's fashionable alternatives. This boom could force a long-overdue reckoning: energy policy must prioritise reliability and affordability alongside emissions, not subordinate everything to net-zero timelines. Countries that cling to ideological purity on coal will watch hyperscalers and their billions flow to more pragmatic jurisdictions, like China.
Australia has choices. We can double down on the delusion, pouring subsidies into unreliable renewables while data centre projects falter or relocate, or we can adopt a mature, all-of-the-above strategy that keeps coal in the mix where it makes engineering and economic sense. Extending plant lives, investing in ultra-supercritical technology, and pursuing genuine baseload alternatives like nuclear (long taboo but increasingly discussed) would provide the firm power backbone these facilities require. The data centre wave isn't a problem to wish away with more wind farms. It's a reality check that exposes how anti-coal mania has left the nation vulnerable precisely when new industries need it most.
The coming years will test whether Australia's policymakers serve the future or yesterday's woke slogans. Data centres don't run on hope and batteries: they run on electrons, 24/7. The generation that ignored that reality for ideological reasons may soon find itself presiding over blackouts, skyrocketing prices, and missed opportunities. The delusion is ending. The coal question is returning. And Australia's energy security may depend on having the courage to answer it honestly.
https://www.macrobusiness.com.au/2026/07/australias-data-centre-delusion/