Australia’s data centre growth proportion now matches the US
A new report was just released by Australia’s grid operator that contains a very, very buried lede.
The Australian Energy Market Operator (AEMO) ‘s quarterly grid report shows that a full 50% of demand growth in the National Electricity Market (NEM) was thanks specifically to data centres.
First: the context. Note the cropped y-axis: this is a 302 megawatt increase in ‘underlying’ demand, but it comes off a base of 25,194 megawatts of total society-wide demand in the National Electricity Market. Q1 2026 is not a step change in total demand increase.
But – demand is still rising. And basically each quarter sees a record high for that quarter pretty consistently now.
What is new here is that data centres are such a large proportion of demand growth, which would have been lower without expansion of these facilities. AEMO has basically never mentioned them as a factor in growth in any previous QED. These are the early stages of a massive shift.
We’re not America*!
A line often used by the data centre lobby group and others is that any comparison between the US and Australia is fundamentally incorrect:
It’s wrong just on its own: through the other side of their mouths the industry is loudly heralding the massive mountains of cash being poured into Australia by Google, Amazon and Microsoft, who’ll all import their attitudes and ideologies into their business plans.
Both the existing and projected future proportion of data centres in demand growth in the US is 50%, easily the highest in the world. That is now already being matched by Australia1, even before the data centre frenzy truly begins. My guess would be that this mostly relates to pre-existing facilities expanding their consumption and filling spare capacity.
Here in Europe, we also hear “but, we’re not like the US!”. Well, Rei Takver from DeSmog snuck into a data centre industry conference and found that gas turbine manufacturers are very, very confident that we’ll see a US-style boom in fossil gas use for data centres. Langley Holdings said:
“I just think the American market is ahead of us [and] the same thing is going to happen here….it just will take a bit longer and it will be a bit harder because more people will be saying, ‘hang on a minute, we don’t want to be burning greenhouse gasses”.
That Australia is not literally identical to the US does not mean there is a clear, noticeable pressure from technology companies to adopt the same attitudes. It is pretty intensely naive to parrot the industry’s lines.
Diverting renewables away from decarbonisation
We can also see from AEMO’s report that rooftop solar growth matched demand growth: meaning half of new rooftop solar served data centre expansion rather than grid decarbonisation. The emissions reductions in this period would’ve been notably lower, without this demand growth.
Across the whole grid, total renewable growth exceeded demand growth. That meant a decrease in fossil fuel use, and accordingly, a decrease in emissions. This is what we want to see. What we don’t want to see is America’s situation: where renewable growth continues, but it’s less than demand growth, meaning fossil fuels rise:
What purpose does the additional data center load serve? Who can we ask? How can they prove it? We know what it means for power demand to rise to keep people cool in a heatwave, or switch from fossil gas heating to a heat pump. What does it mean for a data centre to expand its energy hunger? Who’s benefiting from that? Why is it considered rude to ask, in the midst of an urgent energy transition and and a crippling fossil fuel crisis?
The percentage growth level for the whole grid is not unusual. But half of demand growth being driven by an industry that isn’t serving climate adaptation (cooling) or action (electrification) is very new indeed.
Don’t worry, our planning pipeline is mostly lies
The AEMO report also shows 5.4 gigawatts of new data centre capacity in the pipeline. Assuming a standard capacity factor for AI-related data centres, that is about 10x the current draw from the grid.
The data centre lobby group launched their own ‘data centre forecast report‘ this week too, seemingly continuing the baffling PR message that their own industry is wracked with fraudulent and false connection requests, so no one should be worrying too much about their growth. Notably, the report is vague and avoids specifying which data centre developers the data centre lobby group thinks are futile and destined to failure.
Of the 21.6 gigawatts in their pipeline (this includes projects that haven’t yet applied for connection, hence why it’s bigger than AEMO’s number), they expect only 1.8 to be operational by 2030. Perhaps some will come online after 2030, but it’s pretty clear they think the vast majority of their planning pipeline basically has no hope of existing.
It is awkward as hell for them, because the lobby group is simultaneously arguing the data centre boom will bring thousands of jobs and billions in cash. Which cannot really be true if their industry is infested with bullshit projects that’ll never come to fruition.
Even if a small fraction of this pipeline ends up operational, data centres will drive far more than 50% of demand growth in the NEM. That just stretches the finish line for Australia’s 82% 2030 target and economy-wide emissions targets even further away, as the renewable industry has to run faster and harder to meet far greater demand than they expected while also replacing coal and gas. In the face of weakened renewable roll-out in Australia, the most likely outcome is simply that coal and gas just generate more, and for many more years past their shut down date.
- The data relate to the “National Electricity Market” on the east coats, which excludes Western Australia and the Northern Territory. The proportion would be a bit smaller if you look at all of Aus, but not much ↩︎