Australia's data center boom threatens to drive power prices higher as facilities consume growing grid capacity.

In July of this year, the Prime Minister stated in a speech that, “We will create a legal obligation for the next generation of large-scale data centres… to build new renewable generation – and firming – to … ensure data centres do not increase power prices for Australians.” While we commend the government’s objectives, as is often the case, the devil will be in the detail. In our view, the following six practical policies are needed to efficiently and effectively achieve the government’s vision.

Utilise Renewable Energy Guarantee of Origin (REGO) certificates from new (yet to achieve Final Investment Decision) renewable energy projects as the compliance mechanism. Data centres should be required to surrender a REGO from a new project for every megawatt-hour (MWh) of electricity they consume. Only by limiting REGO surrender to new projects will the government realise its objectives of preventing rises in electricity prices and emissions.

Large data centres must enter into long-term contracts or Power Purchase Agreements (PPAs) with new renewable generation projects. Some have suggested that data centres should simply acquire and surrender new-project REGOs to meet their obligations. However, for over 20 years, the vast majority of renewable energy projects not owned by an electricity retailer have required a long-term PPA to secure the revenue certainty that equity and banks demand to finance these projects. There is no reason to believe this has changed. Furthermore, to effectively mitigate the power price impact of a data centre, the new renewable energy project must be located in the same state.

Incentivise data centre owners to minimise the gap between when they add extra demand versus when new supply comes online. Currently, data centres tend to be built faster than the industry can construct new wind and solar farms. This gap must be kept to a minimum and cannot become a loophole allowing data centres to avoid paying for matching new supply.

Some have proposed allowing data centres to surrender any type of renewable energy certificate—such as REGOs or Large Generation Certificates (LGCs)—during a transition period until new projects are constructed. This approach, however, presents serious problems. First, it will not reduce power prices or emissions. Purchasing REGOs from a 10-, 20-, or 40-year-old renewable generator does not result in additional megawatt-hours of generation, which is weather-dependent. A data centre could buy three years of REGOs while simultaneously purchasing all its physical electricity from a coal-fired generator. Does this constitute a sensible transitional obligation?

Second, it fails to provide an enhanced price signal to encourage new project development. The supply of generic LGCs and REGOs from existing and already committed projects far exceeds demand, even if all current and future data centre consumption were offset by renewable certificates (see supply-demand analysis at the end of this article). Certificate prices have plummeted from over $40 to less than $5 over the past 18 months due to massive oversupply, rendering them economically insignificant for new projects. Adding data centre demand would do little to lift certificate prices enough to make new projects viable. It is easy to see how cheap certificates from older projects could become a tempting loophole, allowing data centres to claim zero-emissions power and perpetually extend any transition period.

A better solution to the construction time lag is to require data centres to cover any short-term shortfall by surrendering additional new-project REGOs later under a grace period. This mirrors the current Large Renewable Energy Target (LRET) scheme, which allows electricity retailers to recover shortfall penalties within a three-year grace period if they make good on deficits. This mechanism successfully addressed the renewable energy investment drought triggered by former Prime Minister Tony Abbott, which left the market short of renewables in the years following his term. A potential middle ground would complement the make-good requirement with a much more constrained transitional supply of REGOs from renewable projects energised after a specific date, such as 1 January 2026. Green Energy Markets’ analysis indicates that operational and committed projects energised from this date onward will produce sufficient renewable certificates to cover incremental data centre electricity demand through at least 2030. To ensure data centres actually contract with new projects, eligibility for these transitional certificates may need to expire sooner. A more constrained transitional supply would create a new certificate market, providing a far stronger investment signal for new construction than permitting the use of any legacy renewable energy certificate.

Data centres must secure the great majority of their firming requirements from new dispatchable capacity. Data centres will need to purchase electricity capacity contracts or write contracts directly with dispatchable capacity and/or electricity retailers to manage wholesale market price exposure when their contracted renewable energy project is not operating at sufficient capacity. If these contracts are not with new dispatchable capacity, there will be a significant adverse impact on wholesale prices paid by other consumers. Therefore, new dispatchable capacity is required, and it must begin generating before, or at least close to, the data centre’s start of operations. Bringing in new dispatchable capacity four to five years after a data centre begins operating will result in years of higher electricity prices. Fortunately, many advanced Battery Energy Storage System (BESS) development projects are underway across the country, and they are quicker to build than wind or solar farms. If a data centre enters a tolling agreement (PPA) with a large BESS, the battery should have a duration of at least four hours and must be located in the same state to effectively mitigate the data centre’s impact on power prices.

The obligation should apply to new data centre capacity committed after 23 March 2026, when the government published its data centre expectations. On 23 March this year, the federal government published its Expectations of data centres and AI infrastructure developers. This document made it abundantly clear to businesses looking to develop new data centres or expand existing ones that they needed to: “Not place upward pressure on energy prices and should make a positive contribution to Australia’s energy transition. This includes working in coordination with energy regulators and suppliers to secure new and additional clean energy generation and/or storage to offset demand.” From that date onwards, any investor in data centres knew or should have known that committing finance to build additional processing capacity would carry an expectation to procure new renewable energy and firming. It does not matter that the government is unlikely to legislate its data centre electricity procurement obligation until next year; the requirements were published in black and white on 23 March. Prudent investors would have accounted for this prior to committing to build new capacity as part of any decent due diligence process, making any claims of sovereign risk laughable.

Data centres supplied by state-owned generators must still surrender new-project REGOs for their electricity use. Even if one accepts the Queensland Government’s argument that state-owned energy infrastructure is a ‘special case’ that might somehow defy economics to deliver lower power prices without adding new supply, it is not a special case regarding Australia’s climate change obligations. Therefore, if the federal government wishes to achieve its emission goals, it must require data centres to surrender new-project REGOs for their energy use, irrespective of any special arrangements with state government power companies. The only alternative for managing emissions would be to require either the state government power company or the data centre to surrender Australian Carbon Credit Units to offset the additional emissions generated by supplying the data centre’s electricity consumption.

Only by adopting these six recommendations can the government ensure that data centre growth does not come at the expense of higher power prices for other consumers, carbon emissions, and climate change obligations.