Data centers: Powering Australia’s AI boom, part 1

Alexandra Peace:
Hello and welcome to Legal Outlook by Ashurst Perkins Coie. You're listening to Here Today, Here Tomorrow, our podcast mini-series about data centers. This is the first in a two-part episode exploring the Australian market.

I'm Alexandra Peace, a real estate partner specializing in data center acquisition and development, and I'm joined today by Kate Muller and Mike Webb, partners in our projects practice. Kate and Mike advise across the full life cycle of energy and infrastructure projects, including data center projects.
Today, we're going to cover the energy interface for data center developers and operators in Australia. Kate, let's start with you. Data centers aren't new; we've had them for decades, but something has clearly shifted. Before we get into the weeds, what's actually changed, and why is this suddenly a policy, regulatory, and grid planning issue all at the same time?

Kate Muller
Well, thanks for asking that question, Alex. And firstly, just delighted to be here and a guest on this podcast.

We're dealing with a shift in scale. We've seen data center activity and interest increasing really rapidly, given the advent of AI and increased data usage and interpretation across our market.
When we think about that as an “energy system issue”, we're seeing the forecasts for electricity demand for data centers increasing in an exponential way. So at the moment, that electricity demand is around 4.7 terawatt hours, and it's going to increase to 11.8 terawatt hours over a couple of years. So that's a really significant shift in demand.

Australia's been identified as a really attractive market globally for data center investment, and so with that, there's a lot of interest. There's a bit of a “gold rush” in the Australian market at the moment.
There's already a significant number of data centers in Australia and, as you mentioned, you know we've had data centers for a long time.

So why now are we seeing a lot of changes?

In New South Wales alone there's been 22 state significant development applications approved or received by 2025 for capacity of around 3.67 gigawatts. So to ground that in terms our listeners might understand, we're talking about 3 million households’ worth of power.

In Victoria, Ausnet is also looking at about 10 gigawatts’ worth of additional requests for connection.

So the pipeline or the opportunity is seen to be massive. However, as you'd expect, the opportunity doesn't always convert, and some of the issues that we're now seeing in data center connections really replicate what we've been seeing in the energy sector for the energy transition for quite some time now. A lot of the interest, a lot of prospecting, a lot of new market entrants, but just how much of that can convert into reality is a space we're watching with keen interest.

Alexandra Peace
And Mike, any opening reflections from you?

Mike Webb
Also great to be here with you, Alex. Thanks for having us. I heard a really interesting sort of analogy that's sort of relevant for this discussion, which came from Jensen Huang, the CEO of Nvidia, who sort of described the whole AI stack like “a layer of cakes” – with energy in particular and water forming that first layer, your GPUs and chips forming the second, your general sort of infrastructure forming the third LLMs, and then apps on top of that.

But it's really exciting to be here today, talking about that sort of foundational layer of energy. Just to sort of add on to what Kate talked about in terms of the real hype and drive in the in the sector at the moment, Australia in a way, is a little bit fortunate in terms of the timing where we are.

We've got a bit of a “benefit of hindsight” when it comes to looking at data centers. We've obviously seen what's happened in the US and the massive demands that have been placed on the power system, and so we have been able to put some reasonably good policies in place.

The Australian government earlier this year released their five expectations for data centers. The second of which was that data centers need to support the energy transition. There wasn't too much more detail in that but, just last month, the energy ministers – within the NEM – have agreed that all new data center developments need to essentially bring their own power and offset their power demand with renewable energy use.

Alexandra Peace
So it is fair to say that the real shift is that data centers are no longer treated as passive loads. They're being asked to be active participants in the energy system, which is a fundamentally different proposition from simply securing power. What do you think, Kate?

Kate Muller 
It's absolutely right, and if we think about the data centers that have been developed historically, often in urban areas, sort of adjacent to load, securing connection in those sites for those early smaller capacity centers was sort of easy to fit with the existing frameworks. It was something that the network service providers could understand – loads of that size and how they would connect.
We have now moved into projects of a much greater scale with different operating requirements, and so that's putting different types of pressures onto the grid operator, onto the network service providers and on the energy system in a way that we haven't seen before.

So as a result of that, you're absolutely right, there's an expectation that data centers will do their part as large energy users in the energy system, and because they're now connecting into an energy system that was already under significant transition, there's new challenges that the data center operators now are being asked to grapple with, because the system itself is undergoing that transformation.
And so, as a customer in that system, they're not just being told, "Well, here's your connection for a guaranteed capacity. You go sign a contract and you're fine.” They're being told that – actually and they're becoming aware – that the system is more fragile, and as a result of that, they need to participate in that system with more technical capacity and capability to participate in that system more actively.

Alexandra Peace
So to that point, where developers say that they've secured power or have a firm connection, what does that actually mean in the current environment?

Mike Webb
Well, in Australia, securing power for large loads like data centers typically means two things.

Firstly, securing a physical connection into the grid means entering into a connection agreement with the relevant transmission or distribution network services provider.

Secondly, to secure a fixed price of energy supply, developers will also need to enter into a retail contract or a direct hedge with a generator or any other provider.

However, in the current environment, securing a connection agreement and entering into an energy supply or hedge agreement does not guarantee that electricity will be available and supplied at all times. Some data center developers are looking for close to 100% power availability (also known as the five nines, being 99.999% availability). Those data center developers, in order to really secure power, will need to consider some form of on-site backup generation for when the grid is unavailable.

Alexandra Peace
The AEMC is currently overhauling access standards for large loads for package two. What does this practically mean for someone connecting a 200-megawatt data center?

Kate Muller 
Yeah, so the AEMC released its draft determination in March this year, and really, it's quite a technical piece of work that's focusing on the standards that will apply to large inverter-based loads, which is what a data center is.

So the draft rule that's been released proposes a three-tier classification, and depending on which of those tiers you fall within, which is based on capacity, there are different standards or processes that you'll need to go through in order to secure your connection.

So essentially, what data centers are now looking at is performance standards and requirements that they can demonstrate their facility can meet, in order to achieve energization and connection at those capacities.

For data center projects that require a greater than 100 megawatt connection, they're automatically subject to the full suite of these new technical standards. One of those new requirements is looking at disturbance right through. And what that essentially means is:

• When there's an issue in the grid of voltage or frequency problem, how does the data center behave during that system event?
• Can it remain connected?
• Can it continue to draw power within the timeframes that are required for that facility?

So the reason why this has become a focus, as we've said, we've got a fairly fragile grid. We also have shared network infrastructure, so customers are all connected to the same network, whether it's “mums and dads businesses” or essential services.

And the concern here is having looked at what's happened, in the U.S. and some other markets, where data centers have essentially tripped off whole sections of the electricity grid.
So the papers released by the AEMC point to an event in Virginia in July 24, where a data center precinct of 60 data centers essentially disconnected an entire part of the grid. Those failures obviously have very real consequences for people who live in these areas, and for us with a very interconnected grid, it has very genuine knock-on effects for people in terms of their security of power supply, but also the cost of power as well.

So the concept here is that the AEMC is looking to introduce these standards again. For those of us who have worked in the energy markets for some time, the standards look quite familiar to us. The studies and engineering that will be needed in order to receive a connection look very similar to battery projects, for instance. And these standards that are being adopted are being sort of assessed against international best practice. Looking to other markets to inform what that can look like: so Texas, Ireland and Finland, for example.

So, what we would say in this regard is obviously: new rules, new requirements. When we think about things like commissioning of connection points and projects, we've certainly seen that that can become a drawn out process. So it's something that we've observed in the electricity sector that it may cause further delays in securing the offer to connect. But even once you've secured that offer to connect, it may also cause delays in connection as operators need to demonstrate their technical capability to meet the standards that are required.

Alexandra Peace
So back to the infrastructure. Who actually pays for the network infrastructure? My understanding is the policy says “user pays”, but is the current regulatory framework fit-for-purpose when you're talking about staged multi-100 megawatt precincts?

Mike Webb
The short answer is “no”.

Back to those expectations: expectation number two is very clear from the government that data centers need to pay for their grid connection costs. The issue we have is the current regulatory framework is based on a lot smaller and incremental load growth, whereas where we are now is we've got data centers looking to connect sometimes up to a gigawatt worth of load, which is significantly bigger than what the regulatory framework envisages.

And in many instances they're looking to develop that connection on a staged basis, so there's two issues with that:

1. The regulatory framework does provide for a sort of concept of “causer pays”. So when you connect, you should be responsible for the costs of that connection. That's all fine. Those connection costs get paid by the connecting entity. The issue we have with data centers is that their connections are so large they're requiring much bigger network augmentation. So you're not just building a small connection, but you're actually having to build out the network to allow for that data center to connect in – as Kate mentioned – that network is used by all other electricity customers, mums and dads, and other businesses. It's a shared network. So what's happening is a lot of those costs are being required to sit on the network services providers’ regulated asset base, and ultimately get paid for by all electricity customers.

2. The other issue with it is that process to put those costs on the RAB is a long process. You need to go through a whole process called an RIT IT to have those costs approved. This is all at the same time where the data centers are actually saying we're very happy to pay for those costs. So you've got data centers Australia coming out in their submissions to the government's policy framework, saying “yeah, the data centers are happy to fund those costs”.

So it's really just about bringing the regulatory framework up to speed with where we are.

The other issue, as I say, is some of these connections are done on a staged basis. So you've got TransGrid and AusGrid saying they're getting connection requests for 800 megawatts, but only 200 megawatts being used immediately. So who funds the overbuild for that? Because you can't easily just build that out and then add that capacity on a staged basis.
Again, I think the data centers would to an extent be willing to fund a lot of that overbuild, but there is also another element where they may not end up using a whole piece of network overbuild, which they then pay for, and someone else gets the benefit from it.

Kate Muller
Just add that the challenges that we're seeing here is the “user pays” model fitting into a regulatory context that isn't really fit for purpose. Again, it feels exacerbated and large in the context of the data center question – but these are not challenges that are unique to data centers.
And when we're talking about large transmission build outs, our listeners might be familiar with what we're seeing happen with the renewable energy zones being built out and the significant greenfield transmission and distribution infrastructure that's necessary.

So there's a lot of work that's been going into the commercial models for this kind of infrastructure, and a lot of questions being asked about the appropriate way of funding that infrastructure, and how much of that sits with the private sector, the proponents that are connecting to that infrastructure, and how much of that is for a broader public benefit and belongs on a regulated asset base and subject to the RIT. So, I think it's not a unique problem to data centers. It's something that's being grappled with in the broader context of the energy transition as well.
So, if we think about the renewable energy zone buildouts, or we think about these new industrial precincts that are being energized – there's a lot of work going into these commercial models, and a lot of thought going into the regulation that sits around them to try and facilitate more private investment and to remove that burden from mums and dads on their electricity bills.

Mike Webb
I think that's a great point, and I'd add (or almost just echo) that you almost want to see these data centers potentially being developed in clusters in precincts, where you might be able to coordinate the supply of all of these services. It's not just energy; they also need water – whether that's recycled or whatever it is, they need transport. They need the fiber connection, so I think there is a real opportunity to coordinate that into precinct developments.

Alexandra Peace
Thank you, Mike, and thank you, Kate. This brings us to the end of part one of this discussion.
Thanks for listening to this episode of Here Today, Here Tomorrow, our Ashurst Perkins Coie podcast mini-series about data centers.

You can find part two of this Australian discussion on Apple Podcasts, Spotify, or wherever you get your podcasts. And that's where you'll find all the previous episodes in this data center series too.
Stay subscribed to this podcast to make sure you don't miss any of our future episodes.

Until next time, thank you for listening, and goodbye for now.