How do we power the data centre boom while also quitting coal? There are three options
Australia’s proposed $150 billion data centre boom comes with a substantial energy bill. Who pays will depend on whether these facilities help build the new renewable supply they need, or compete for existing electricity with households and businesses.
Carbon Zero Initiative has been working on this issue since August 2025. In that time, we have seen proposed investment triple to $150 billion. Communities are asking reasonable questions about what this means for their power bills, water supplies and local infrastructure, and how much of the promised investment will actually stay in Australia.
The federal government’s proposed data centre standards are an opportunity to answer those questions.
Our submission to the Office of AI focuses on energy: how to turn demand from data centres into investment in new renewables, and what obligations are needed to protect other electricity users.
Let’s begin with the big numbers. According to Oxford Economics, only about one-third to one-fifth of that $150 billion is actually realised investment in Australia. Staggeringly, the rest goes primarily toward the ultra-powerful, ultra-energy-hungry computer chips designed by companies like NVIDIA and built by Taiwan Semiconductor.
Power consumption predictions for new data centres also vary, but the Australian Energy Market Operator provides a good starting point. It predicts 14% of Australia’s energy will be used by data centres in 2036, up from about 5% today. That’s more than all households in Sydney and Melbourne combined.
Interestingly, the large gigawatt (GW) numbers thrown around in the data centre world need to be immediately halved. Data centres seemingly reserve twice the capacity in their connection requests as they actually use.
This is likely a combination of ensuring physical redundancy for their power supply and accounting for potential future expansion. This oddity is confirmed by recent studies from the International Energy Agency and University of California, Berkeley.
The final numbers then are 4 GW flat load by 2036 (AEMO, ESOO, 2026) which requires approximately 10 GW of solar, 6 GW of wind and 16 gigawatt-hours (GWh) of batteries in order to be powered by new renewables (CZI calculation).
Importantly for planners and policy-makers, the load doesn’t hit all at once, it arrives in four equal stages, about two years apart.
Baringa (2026) and AMEC (2024) both modelled how un-matched new demand of this size would affect energy prices, arriving at similar conclusions. Wholesale electricity prices would rise by about 20 to 30 per cent absent new renewables joining the grid.
Others model what happens if the 4 GW arrives all at once and much sooner. Unsuprisingly, the price jump is much higher (Endgae Analytics).
So can we do this? Are there enough projects to meet this demand whilst also switching off coal, gas and electrifying industry and transport?
The Clean Energy Regulator (CER) estimates that there is an existing pipeline of 8.7 GW of committed projects, with a ‘probable pipeline’ of up to 32 GW of combined wind and solar capacity.
We must accelerate this pipeline of probable projects; otherwise, power price rises are inevitable. If we want to realise data centre investment, electrify transport and industry and turn off coal and gas we need to grow the pipeline substantially.
How do we get these two industries to work together to build out energy and digital infrastructure? There are basically three options, all with pros and cons.
The first option is to let the market sort it out. Some data centre developers will voluntarily enter into power purchase agreements (PPAs) with new renewable energy, others will not. Those that don’t will push up power prices, but this will incentivise new renewable energy to join the grid eventually.
Whilst that might work in practice, it is not a good outcome for consumers or businesses.
Another option would be to place guardrails around the data centre industry with a renewable energy obligation. We followed the Office of AI’s language and called this a data centre renewable energy obligation (DCREO).
At its simplest, this is a renewable energy target that applies to every data centre in Australia over 30 MW, including exisitng facilities or those under construction, requiring them to either enter PPAs with renewables that didn’t exist when the data centre started operating; or to purchase offsets.
The offsets would also need to come from relatively new renewable projects (we suggest 1 January 2024, in our submission); otherwise, the data centre could free-ride on the deep pool of existing cheap offsets while pushing up everyone else’s bills.
A compliance function must also be built into the certificate obligation to prevent indefinite free-riding. This PPA and offset approach is likely to be most favoured by data centers and the renewables industry because it rewards good actors and provides them with some flexibility.
The third and final option is for the government to step in. This could be through contracts for difference, a special purpose investment vehicle or the Clean Energy Finance Corporation making loans to renewable energy companies to accelerate construction.
This helps build new energy, but doesn’t actually create the conditions for data centres to buy it, so a compliance function would still needed here as well.
On balance, the second option seems most likely to succeed as it seemingly has industry support from data centre developers and the renewables industry, but the devil will be in the details.
The government has flagged that it will introduce bills to parliament on this in November 2026. In a month! That seems ambitious given the presumed need for the Office of AI to sift through potentially thousands of submissions it has received, incorporate feedback, write the bill with ministers and then consult on it with industry, civil society and cabinet.
We understand the urgency. Communities are demanding answers from their elected representatives about whether the AI investment boom means their bills will rise.
But rushing the bills through may not solve the problem if it lets data centres free ride on the grid that Australian businesses and householders have invested in.
This Op-Ed by Jack Redpath is featured in Renew Economy.