Andrew L. Urban
Australia’s obstinately zealous Climate Change and Energy Minister Chris Bowen is speeding along Transition Highway 1 (aka Route 666), ignoring the roadside warning signs, his foot on the accelerator. The Detail Devil awaits …
Bowen’s policy EV will need lotsa airbags when it crashes into the road block of the insatiable electricity needs of AI. And pretty soon. There is no sign of preparing for this crucial detail in his plans. The explosive growth in artificial intelligence (AI) and data centres is introducing a new variable: surging electricity demand that could strain the grid and challenge this trajectory. Globally, data centres consumed 415 terawatt-hours (TWh) in 2024 (1.5% of the entire world’s electricity), projected to double by 2030 due to AI.
In Australia, data centres already account for around 6% of electricity use, also expected to double by 2030, adding up to 4.9 gigawatts (GW) of demand—equivalent to 2-3 times official forecasts from the Australian Energy Market Operator (AEMO). AI-specific growth could hit 25% annually for a decade (ie doubling every four years), potentially making Australia’s data centre consumption exceed Japan’s current total by 2030. This could dramatically slow the pace, increase the costs, and tilt the energy mix toward more fossil fuels (especially gas) in the short-to-medium term if grid constraints persist.
A move to favouring fossil fuels (particularly gas) is probable in the 2025-2030 window to bridge reliability gaps.
Or is AI even more consequential? Is AI a dagger pointed at the heart of current climate change policies – especially in transition-aggressive Australia? Video killed the radio star … will AI kill the climate alarm star?
The renewables rush—fuelled by falling solar/wind costs and coal retirements—faces headwinds from AI’s insatiable, 24/7 power needs. AI data centres require constant baseload power, which intermittent renewables (solar at some 30% capacity factor, wind at about 40%) can’t reliably provide without massive storage and transmission upgrades. Without these, grids risk blackouts, higher prices, and delayed ‘decarbonization’ – an issue for those who demonise CO2. Fossil fuels, especially gas, offer quick dispatchability and could fill 40-60% of data centre needs globally by 2030 if renewables lag. In Australia, this could mean extending coal/gas plant lives or building new gas peakers (power plants that generally run only when there is high demand), countering the 82% renewables goal.
Grid bottlenecks (e.g., transmission delays) could further force fossil reliance. Oh, and speaking of higher prices… not the ones we’re going through (37%) but the ones coming from 2025 to 2027. Higher wholesale prices (up another 20-50% short-term) from volatility and subsidies for renewables ($1B+ according to ARENA, the Australian Renewable Energy Agency). “Subsidies for renewables” is Bowen-speak for taxpayer-funded grants. Think of ARENA as Australia’s public “venture capital” fund for renewable energy. It deliberately takes on projects that are too risky or too early-stage for banks or private investors, with the goal of de-risking them so the private sector can then scale them cheaply. Do you realise you are a major, obligated, unsecured (and probably unwilling) investor – but not beneficiary – in renewables?
In summary, AI’s demands expose renewables’ growing pains, likely slowing Australia’s transition by 5-10 years and boosting fossil fuels, at least temporarily, for reliability. Success hinges on urgent grid reforms and hybrid solutions—failure risks economic isolation in the AI era. This aligns with IEA (International Energy Agency) warnings: diverse sources (renewables + gas) will meet AI needs, but physics demands planning over ideology.
So, not so fast, Minister …
And yes, if you’re wondering, AI helped with the research for this article.
Andrew L. Urban is the author of Climate Alarm Reality Check (Wilkinson Publishing).