Andrew Charlton (Parramatta, Australian Labor Party, Cabinet Secretary) | Hansard source

I thank the member for Warringah for this matter of public importance, and I thank her and many members of the crossbench for the way they've engaged in this issue, balancing both the risks and the opportunities. The question of how Australia approaches data centres is an important one for our nation. It's important in part because of what's inside those data centres: artificial intelligence, which is going to be one of the most consequential technologies in our economy over the next several decades. In the 19th century, economic power rested on coal and in the 20th century on oil, and in the 21st century it will rest on computation. AI is the issue; data centres are simply where the AI happens.

The Albanese Labor government is determined to ensure that all Australians benefit from AI. To get those benefits we need to shape and shepherd AI to make sure that it works for Australians and not the other way around. Late last year the government released its National AI Plan. It set out a simple proposition that the role of government is neither to stand back and hope nor to stand in the way; our role is to ensure that AI develops in line with Australian values, Australian laws and Australian interests. As part of that plan, we've set out a set of clear expectations for the companies building the physical foundation of the AI economy: data centres. Those expectations include what I call a triple lock on energy. There'll be members of this parliament who want to pretend that we can stop or ignore AI. I think that's irresponsible. AI is here. The arrow is loose from the bow. Our job is to make sure we benefit from it.

It's important to note that the AI boom is no small thing. The investment now flowing into data centres is on a scale that is difficult to overstate. Moody's expects global data centre investment to reach $3 trillion in the next five years. For Australia, this boom could ultimately prove to be bigger and more consequential than the mining boom—a boom that shaped our prosperity for a generation. The question isn't whether the boom is coming; the question is: on what terms do we let it in? The government's approach is not to be boosters of AI and not to be naysayers. We want to be pragmatic, and we need to be clear-eyed about both the big risks and the big opportunities that flow from this boom.

Let me be clear about the risks first, because they are real and Australians can see them. First, let's talk about energy. Around the world, data centres are pushing up prices in countries where these centres have not been well managed. In the United States, it's already happening. In the PJM grid, one of the largest in the country, the Independent Market Monitor attributed more than half of one year's surge in capacity prices to data centres—more than US$9 billion in additional power prices billed straight back to households. We cannot let this happen in Australia, and we need to act early to make sure that it doesn't. Today, data centres use around two per cent of the electricity in our National Electricity Market. That sounds modest, but the trajectory is steep and, if demand arrives faster than new supply, the consequence is obvious: prices rise and households feel it.

There are also concerns about water. Nationally, the volume is still small. Data centres use around one-one-hundredth of what mining uses, but that use can be locally acute, and some communities overseas found that out when taps ran dry during construction of data centres near them. There are also community concerns about the noise of data centre clusters, their visual impact and industrial sites built close to homes. These are real concerns and they deserve real answers. We've seen it here in Sydney. A single proposed data centre in suburban Sydney drew a record 374 objections and just nine submissions in support. Residents were worried about noise, diesel generators and a facility that is barely 160 metres from a local primary school.

These risks from data centres are not theoretical. They are materialising around the world, and almost every country that was slow to address the data centre boom has had problems caused by it. The United States let the market lead and is now improvising after-the-fact fixes amid a huge public backlash. Ireland let data centres pass 10 per cent of national energy demand before the grid around Dublin seized up and new connections were frozen. Singapore has had to halt approvals. The pattern around the world is unmistakeable. The countries that were slow to manage the costs were the ones that faced the worst consequences, so that's exactly what our approach in Australia addresses.

Under the National AI plan 2025, we've released our expectations of data centres and AI infrastructure developers. These are not red tape. They are the foundation of the sector's social licence to operate here. They are a plain statement of what we expect in return for access to our grid, our land and our market. They're framed as expectations, as the honourable member pointed out, because they cover areas of responsibility that intersect with state, territory and local governments. For that reason, as a federal government, our approach has been to outline our expectations first, which we've done, and then work with other jurisdictions to create alignment and certainty to make these expectations binding across the Commonwealth. That is the path we have laid out and the path we are making progress on.

Let me lay out what these five expectations do. The first expectation is the national interest. We expect that this investment will build genuine Australian industrial and technological capacity, advancing our sovereignty and security, not simply adding floor space and consuming our resources. The second expectation is energy, and it's at the heart of this framework. Here we are applying our triple lock. The first component of that lock is that we require data centres to bring their own renewable energy supply, underwriting fresh generation rather than drawing down power that households and businesses are relying on. Second, we expect them to pay their full share of grid connection to cover the cost of the poles, wires and transformers they require and make sure that those costs are not passed on to people's bills. And we expect them to be demand flexible, working with market operators to dial their consumption up and down so they strengthen the grid rather than strain it. These three things—bring your own supply, cover your network costs and be demand flexible—are the triple lock. These are the things that enable us to look Australians in the eye and say that the growth of data centres will not push up your power prices. Energy ministers across the Commonwealth, states and territories are now working together to embed it.

The third expectation is water. We expect operators to use efficient, modern cooling systems to draw on recycled and non-potable sources wherever they can.

The fourth expectation is workforce and skills. We expect this build-out to invest in Australians through apprenticeships, training and partnerships with our colleges and universities, so the boom leaves behind a deeper pool of local capability than it found.

And the fifth expectation is innovation. We expect the global operators to make their computing power available to Australian start-ups and to partner with Australian researchers, so the benefits of this build out flow to Australian ideas, not merely Australian real estate.

This last expectation matters a lot, because it goes to the question underneath all of this: why are we all doing this? The AI economy is coming, whether we are ready or not. What is not settled is our place in that economy. We can be a country that rents intelligence by the month from foreigners on terms set in another hemisphere, or we can be a country that helps build it, owns the infrastructure, trains the people and shapes and regulates the technology to our own values and our own interests.

The choice is not whether the boom arrives; it is whether Australians end up as tenants in someone else's digital future or owners of our own. At its heart, whether we have AI in Australia or not is an issue of sovereignty. When our government activities, our research and our AI models run on Australian soil, we are not a tenant in someone else's economy. And, in a contested region, that is not a luxury. It is about national resilience.

That's why this debate matters so much. We must be clear eyed about the risks and the opportunities and make sure that data centres and artificial intelligence benefit all Australians.

See this speech in context