Prime Minister Anthony Albanese announced plans this week to bring in standards for artificial intelligence and data centres. Under the proposal, operators would have to underwrite new electricity generation in line with how much power they use, pay their share of the cost of running the electricity network, and cover any extra water infrastructure their sites need. The Australian legislation is expected to pass parliament in early 2027.
The concern behind the move is that surging demand from data centres pushes up wholesale electricity prices for everyone else, CommBank Sustainable Economist John Oh says.
It’s a trend that’s already showing up in the US, Oh said in a research note. In high-demand US regions like Virginia, total electricity prices rose 76% over the year in the first quarter of 2026, he said.
Why underwriting power is more than a green tick
Australia’s proposed new rules would force data centre operators to help build new renewable power generation capacity, rather than just buy it, Oh said.
“Although current data centre operators can obtain green power through purchasing large-scale generation certificates (LGCs), the requirement to underwrite new electricity generation is a step up in commitment,” Oh said.
The new rules would likely require onsite renewable generation installation or indirect funding of renewable projects through power purchase agreements (PPAs), he said.
In practice, that would mean committing to pay for power well ahead of time. Those contracts would usually be locked in before a renewable project gets its final go-ahead, so a data centre would need to sign up for new supply years before it actually starts generating, Oh said.
Building power takes longer than building data centres
Timing supply and demand could be tricky. Oh noted onshore wind and solar projects in New South Wales and Victoria take between three and five years to build and approve, according to research firm Rystad. But data centres can be built in as little as two years or as many as seven, depending on their size.
“The key risk is that data centre demand arrives before generation comes online,” Oh said. If a site switches on before the power it has paid for is ready, that extra demand still has to be met from the existing grid, which is exactly the pressure the rules are trying to avoid.
Why batteries might matter more than new power plants
The rules proposed by the government focus on power generation, but energy storage might end up doing more of the heavy lifting, Oh said.
Electricity is most expensive during the evening peak, and that is the hardest time to guarantee renewable supply.
Solar is strongest in the middle of the day, and wind in the evening cannot be relied on. Batteries help bridge that gap by charging when power is cheap and plentiful, then releasing it when demand and prices climb.
“Storage requirement is likely to better shield the grid from higher data centre demand,” Oh said.
How much extra power we are talking about
The scale of the potential jump in demand is significant. The Australian Energy Market Operator (AEMO) expects data centre electricity use to reach 15.6 terawatt hours in 2030-31, up from 4.7 terawatt hours now. That would lift data centre demand from about 2.6% of national electricity demand to 6.3% in the space of a few years.
AEMO is due to publish its next reliability assessment in the month ahead, with fresh forecasts for how much power the country will use and where it will come from. Oh says the market will be watching one issue closely, the planned exit of several major gas and coal power plants in 2028-29, and whether new supply arrives in time to replace them.
This article is based on John Oh's CBA Commodity Daily, Unpacking Australia's data centre standards, 17 July 2026.
