CUMBO, Dr Bronwyn, Lecturer, University of Technology Sydney REY, Dr Jo, Private capacity WRIGHT, Dr Ian, Associate Professor, School of Science, Western Sydney University [15:24] CHAIR: Welcome. Before we go to opening statements, do you have any comments to make on the capacity in which you appear? Dr Wright : I am formerly of the water industry, and I do peer reviewed science to help manage tricky water problems. Dr Rey : Hello. Warami. I'm a Dharug community member. I'm also a researcher at a university, but I come today as a private individual. CHAIR: Thank you. Let's kick off with you, Dr Cumbo. Dr Cumbo : Thank you for the opportunity to appear today. I've been researching Australia's evolving data centre industry over the last two to three years. The federal government have clearly outlined, in their national expectations and recent consultation paper, that they want to attract good data centres that align with Australia's national interest, which is a wonderful development. I just want to ask two questions to expand this discussion and our idea of what a good data centre looks like. Firstly, can a good data centre be a trustworthy one? This includes things like being open about resource use, emissions and independent reporting and taking steps to be proactively sustainable as the tech industry and the various infrastructures evolve. How about their relationship with communities? Trust is about building relationships and partnerships and reframing how communities are perceived by both the industry and government. We can't simply see the communities as a hurdle to the process; they're potential partners in innovation and stewards of their local community and environment. We need them on board so that this industry is sustainable. Secondly, how can we sensibly assess if a good data centre will be worthwhile? We're talking about the data centres that don't exist yet. A large part of this industry is built upon the intoxicating potential of generative AI and what it might bring our society. By building this infrastructure we're increasing our chances of this potential being realised here in Australia, but there's no guarantee it will live up to the expectations. Uncertainty, hope and optimism are driving much of the industry's momentum. What we do know with certainty is that data centres come with material, social, economic and environmental costs. So how does a good data centre balance these known costs against potential, and deeply uncertain, future benefits? And who gets to decide the level of risk we're prepared to accept? CHAIR: Thank you. Dr Wright? Dr Wright : Thanks for the opportunity to be here. I've learnt a lot already today. My concerns, firstly, are that data centres are a newish, very thirsty industry that is expanding very quickly and will potentially have a gargantuan thirst—I don't know the metric equivalent of that—for cooling water and will often use the community's drinking water as cooling water, across many water stressed communities in Australia. I haven't met a community in Australia that isn't water stressed. We are water stressed in Sydney. We're at our limit, now, but the dams are nearly full. How full? About one and a half terms of government. They're at 95 per cent at the moment. So we have, at the moment, a false sense of water abundance. We're currently at 93.9 per cent. Perth is at just under 40 per cent. The New South Wales government developed a Greater Sydney metro strategy that made it very clear—in August 2022—that we are now at our current sustainable supply level of roughly 550 gigalitres a year. It's sobering reading because it's about half the level from the previous 112 years of inflow data. If you look at Perth and south-western WA, the decline is even more than that. We've got more people and now a brand-new industry. At the moment, yes, the dams are full, but I think it's a false confidence. Data centres will add to our collective thirst. I'm talking about communities in most of Australia going into a deeper drought more quickly for longer. I don't completely agree with everything I heard from Mr Plowman from Sydney Water earlier, but I think a lot of the costs will be pushed on the community. In the millennium drought and in the recent Black Summer drought, we instituted in Sydney the toughest water restrictions for any large utility. For example, people in often bushfire prone situations couldn't use a hose. Why are they not using wastewater now? I've only really got a list of questions for you. I think it comes down to cost. I recently wrote an article for The Conversation, trying to pull together and answer some questions about water use, and it is so hard to get any information on this. I can't wait for your report to come out; I look forward to it. One of my solutions would be something like the Commonwealth's National Pollutant Inventory, which has fantastic information on large polluters right around the country. We could have a national data centre inventory of water use, power use, contaminants in wastewater et cetera. Dr Rey : Warami. I'm a Dharug Ngurra community member and also an academic researcher. I'm going to present through a human lens to start off with and then move into the Indigenous lens. Futures dominated by AI systems that are not appropriately managed are doing harm to individuals, states, nations politically as well as in health, finance, and personal identity safety and misappropriation. This represents a total willingness to gamble not only on the future of this continent but on that of the planet. We are already facing shocking consequences from human-induced climate catastrophes. Allowing unfettered growth of AI and data centres shows that human beings cannot be trusted to produce sustainable futures. So the questions become: What are the motivations, and who are the motivators who want unfettered access and control? What are the consequences going to be? The race between China and the US to control the AI technosphere has recently been commented on in the media—some of the unknowns and questions they raise. AI and data centres rely on critical infrastructure, and this was discussed just earlier, including subsea cables and terrestrial fibre networks. It is common knowledge that China has been trawling Australian waters and there are concerns they are mapping the subsea cables. Thus, how can governments assure us that AI data centres are not going to be subject to Chinese or other interference? Last night on Four Corners, the documentary, Asleep at the Wheel, showed how Chinese EV vehicles such as BYD can be infiltrated and controlled by China to a point where a driver's or owner's data can be tapped, elements can be manipulated, and personal security, as well as national security systems, can be maliciously tampered with. This is not a separate issue. It raises the question of technosecurity management and the lack of sufficient security systems being ready and in place, and I think you were talking about this earlier with the water and the pollution in the water. So what initiatives have been put in place to protect urban residents and natural landscapes upon which we rely from malicious intent? To what extent are they protected? In relation to Dharug Ngurra, we are here today. I'd like to pay my respects to elders past, present and emerging who may be here today, and recognise that they always have been and always will be caring for country. In relation to Dharug Ngurra, AI causing negative impacts, specifically in the Lane Cove River vicinity, include: sound pollution beside the river and a national park impacting marine life, bird life and human neighbours; water loss and pollution; and fire risk. I'm not sure if fire risk has been brought up yet today, but being beside or near national parks is a fire-prone location. What systems are in place to prevent them from causing a fire risk or, alternatively, causing disasters if they are struck by wildfires in the valley? My research is currently to do with cultural burning, in contrast to mainstream hazard-reduction burns. Another risk is aircraft risk. Currently, air traffic runs along the Lane Cove River valley every few minutes during the day, and anyone who is in the Lane Cove National Park on a regular basis knows how often the planes come over. It does so because there are fewer human beings there, thus it is considered a safety corridor. However, having data centres along this safety corridor nullifies this current safety strategy, should an accident take place. From an Indigenous perspective specifically, beyond the physical risks and the threats to current political and other human systems, the following concerns are also significant to traditional custodians, as articulated by Indigenous scholars Professor Bronwyn Carlson and Dr Tamika Worrell. AI data centres take extensively from country without much benefit to the communities affected. AI and particularly generative AI impact Indigenous peoples through both the theft of knowledges and amalgamating diverse and unique Indigenous nations into one, when we know that there are 250 different nations across the continent. AI is not capable of centring relationships—and we heard a little bit about that earlier—and relationality is at the heart of 65,000 years of continuous culture on this continent. There are more than 5,000 known archaeological sites across Dharug Ngurra, and just within this river valley, between Sugarloaf Point and the Figtree Bridge, there are around 65 recorded archaeological sites. Caring for country so country can care for us is critical to continuity. Indigenous people are not appropriately consulted, yet we feel the biggest brunt of the impact in racist-centric environments—for example, social media et cetera—as well as so-called developments affecting and showing no relationality and no care for the continuity of the systems on which we depend: clean water, clean air and clean foods. As quoted in today's Sydney Morning Herald—and this is my conclusion—by David Swan, in the article '"I'd love to see more of it": Telstra backs data centre boom amid backlash', Australia already has 162 data centres agreed upon and another 130 in the works. Based on all the considerations I've outlined above, how is it credible that any responsible government would wish to offer such a future that sees no known parameters, based upon a gamble that all will be well? Do they really think we are so stupid? Gambling our security and our future generations' security on something that someone tells us, 'Don't worry; it won't hurt you'—it's too much. Thank you. CHAIR: Thank you. If only we got clapped every time we presented in the Senate! If you were in Canberra, you'd be told that was out of order! But we'll leave it. Senator Ghosh, go ahead. Senator GHOSH: [inaudible] evaluation of data centres in a principled way, or is there a framework or a series of tests that we can apply when considering whether data centres in particular locations should go forward or, as a more general proposition, should be permitted in Australia, given those trade-offs? Are you aware of any frameworks or similar sets of principles that you think are valuable in making that evaluation? CHAIR: Sorry, we missed the beginning, Senator Ghosh. Senator GHOSH: Sorry. Would you like me to start again? I can probably do it quickly. CHAIR: Okay. Senator GHOSH: What are the principles or the frameworks that we should use to evaluate whether data centres should go forward either in particular locations or more generally? CHAIR: We got that. Dr Cumbo : Was that directed at me? CHAIR: Well, any of you, I think, but let's start with you, Dr Cumbo. Dr Cumbo : Okay. I think, within the expectations, it's a great start, and they're there. I think the details of what that means definitely need further unpacking—and the data. As Dr Wright mentioned, it's very hard to get information about actual water use, actual energy use and actual emissions. There's so much uncertainty around the industry in general, and I think that really works in the industry's favour at the moment. All infrastructure developers are trying to get certainty. They really require it to be able to build, for instance, the recycled water that we talked about in the northern region of Sydney, where what's going to happen with that is kind of dependent on the industry coming forward, which is not a great position to be in. In terms of where they should go. I think we need to think about what people are saying. Location is really important—the community. What are their values around this? What are their actual benefits? More broadly, for the public, how is this industry in this location shaping things like water security, trust in water governance, energy, our energy transition and climate change—these core values that we have—and also the everyday experience of living in that neighbourhood? Lane Cove is a great example currently. There are going to be a lot of sensory impacts, and the location of these data centres is kind of a game, because it's an industrial area and they're technically allowed to build there with the current rules. So that's what we're seeing happen. It's not a considered planning process. So I think there needs to be greater coordinated and strategic planning around where these pieces of digital infrastructure are placed, and we should also be seeing them not just as individual warehouses but as visible nodes of what is very much a global digital infrastructure that interact and completely reshape water and energy infrastructure. If that's happening, how can we align that with the REZ or water-recycling plants? This requires time and planning. I think this push and this idea that this is a race and we need to catch up now is part of the problem with the industry, because it's pushing and making decision-making very tricky, in terms of the actual realities of planning. CHAIR: And, again, there is this kind of fragmented approach to both approvals and the understanding of impacts. Dr Cumbo : I don't think that's an intentional thing. The industry caught everyone on the hop, so what we've seen is a lot of developers just buying land and putting proposals in, without a collective and cumulative understanding of what's going on. For instance, Sydney Water and energy companies are seeing this case by case. They're getting it proposal by proposal, without having a clear overview of what's happening. It's really ramped up in the last 12 months. So they're raising concerns, but, again, it's concerns around uncertainty. So how do we manage this great unknown? CHAIR: Dr Wright, have you got anything to add to that? Dr Wright : No. I think that was well worded. But I think we're missing a great opportunity. We do have data centres. What's the monitoring? What's the adaptive management? What can we learn from that? I think that was really well worded, and they were some of the questions I was looking for when I was looking at what is constructed now, what's being planned and what the problems have been—the energy use, the water use and how much recycling. So many of them have their own fossil fuel power, so what are the carbon emissions going to be? But the opportunity is to study that now. I'm sorry to say it, but we should use the communities that already have it. We should learn from that. CHAIR: Senator Ghosh, this is just a follow up to your question. Then I'll give it back to you. Actually, Senator Ghosh asked this question to some other witnesses earlier today, and I'm just thinking from an academic perspective. Are there types of frameworks that are now being used or adopted elsewhere in the world that we could learn from, as you're suggesting, Dr Wright? Has any of your work gone to any of those comparable countries? Dr Cumbo : I could take that on notice. I think the problem with the frameworks is that they tend to focus on something like community social licence or water, so they remain quite fragmented. For instance, there's a lot of work around the energy transition and how the industry could potentially accelerate it through investment, so it's ignoring a whole lot of other factors. So how can we think about this holistically? I haven't seen any frameworks. I'm hoping there are some somewhere in Scandinavia, but it's tricky to know. CHAIR: There was some suggestion of something similar in Norway. If you're keen to take that on notice, that would be helpful for us. Back to you, Senator Ghosh. Senator GHOSH: You face a problem—that you're trying to set rules of general application from a legislative or even a regulatory perspective, but those rules need to accommodate the changes in technology and changes in the way some of these things are rolled out quite quickly. For instance, one of the ways that's been proposed in the artificial intelligence space more generally is an AI safety institute. Do you have any ideas about how you would deal with that in relation to a physical infrastructure topic like data centres? CHAIR: So an overarching agency that can map and hold to account—an accountability measure; is that what you're referring to? Senator GHOSH: That might be the answer, but how do you deal with the problem of technology moving quicker than regulation can in this space? Dr Wright : I would like to see more recycled water used for data centres, but one of the problems is: which guideline do you use? At the moment, each state and territory either doesn't have them or has their own guidelines. I think Victoria has what they call 'class A recycled water'. I think Queensland is quite well equipped, with the South-East Queensland recycled water grid. I would like to see some national standards on what would be appropriate. I'd like to hear from the data industry: what recycled water can you take? It is not straightforward treating sewage. It's very tricky. It's an expensive and tricky process. But I'd like to see standards for the industry to aim for. And I think that would also help the water utility to know how they can deliver that wastewater. That seems to be a pretty big block at the moment, a speed bump that's diminishing the ability to cool data centres with treated wastewater. CHAIR: Sydney Water did tell us just previously that it would be much more expensive to use the recycled water. Dr Wright : Absolutely, and they lose money on recycled water. The purple pipe system that they're rightly proud of at Rouse Hill—I think it's piped. That's highly treated recycled sewage and other wastewater. They lose money on every litre that they produce. Sydney Water are highly regulated by the Department of Health EPA but also IPART, the Independent Pricing and Regulatory Tribunal. They make recycled water cheaper, but it costs more, and that has diminished the expansion of that scheme into new suburbs. When Mr Plowman was talking—I can understand completely why a data centre would be signing up for mains water, especially at mates' rates for a big commercial enterprise. I could see why they'd do that rather than investing. The latest recycled water scheme they built was out at St Marys. It's offline at the moment, but that was something in the order of $200 million. I've taken university classes there. It's very complex, and it's very difficult to achieve that. I can see why our data centre would quickly sign up for mains water, and with enthusiastic state premiers and others—'The mains water, yeah, just hook it up!' It's probably three years before we're going to have a big drought here. CHAIR: Is anybody else wanting to respond to Senator Ghosh's question? Dr Rey : I'd just like to say that the conversation tends to be from the top level down in terms of national, and then everything gets determined from that way down. Whereas, if it was done culturally, it would be done from the ground up. Local communities would have input and receive information, just like we're doing today. We're actually here locally and we're having these conversations this way. For me, I feel this is the right way about it. As for responding to Senator Ghosh's question, the emphasis and the input should be coming from the ground up. We should be having informed discussions before anything else goes forward. At the moment, we've got all these things going through planning, but there haven't been the conversations. There haven't been the questions around what risk factors are associated with not just water but all the national park animals—all the various species that are under threat that we all care for. This area is totally linked to the national park. We've got another one that's being talked about over on Alma Road in Macquarie Park. There are already five data centres that I'm aware of in Ryde, in Macquarie business park. The one that's being proposed at the moment is right next to the freeway, and then immediately on the other side of that is national park. If there is a fire, what happens? What's the consequence of a fire in a data centre? How big is that? Do we have any information about how they're going to respond to fire, for example? Another one was being proposed, and it was literally on the edge of the national park, so National Park could speak to that. The impacts on these really rare species—we're in a city of six million people or thereabouts, and we're so dependent, for our mental health and our wellbeing, on these green spaces. Yet we're willing to threaten them by putting data centres—and other things, but let's focus on data centres—right next to them. Imagine if there were a plane crash along the Lane Cove valley and it struck. What would happen? What would the disasters be? It's the same with the fire. If a megafire goes through—and I've literally been looking at hazard reduction this morning—their ability to manage fire risk in national parks is really worrying. When the national park sits next to a data centre, all the preliminaries need to be set in place first and then the planning. Thank you. Sorry. CHAIR: No, all good. Senator Gosh, I'm going to rotate the call. If you've got other questions, I'll come back to you. I'm going to go to Senator Henderson. Senator HENDERSON: I'm okay for questions. CHAIR: Okay. Senator Pocock. Senator DAVID POCOCK: Thank you very much for your time here today. Dr Wright, I might ask you to expand a little bit on some of your concerns with water. As you said, Sydney Water are feeling pretty comfortable at the moment. The dam is full. They said that they expect data centres to go from one per cent of water usage up to five per cent by the end of the decade, and they said, in a drought with water restrictions, they would have to do something. It's very unclear what they could do. I'm interested in your views, given the scarcity of water, on what we should be thinking about. Dr Wright : Firstly, I question that five per cent figure because I've heard in the media Sydney Water quoting up to a quarter of supply—which, if my maths is okay, is closer to 25 per cent. It probably also speaks to the uncertainties involved. I accept that it might be five. It might be 25. We don't know. But, given that we are at our sustainable supply level— Senator DAVID POCOCK: Could you expand on sustainable supply. CHAIR: What does that mean? Dr Wright : That's based on the state government metropolitan Sydney water strategy, which was done not that long after Black Summer. In August 2022, that was released. It took my breath away, for a government document, because it was well written and it didn't shy away from 'guys, we've got to watch our water level'. We're getting about half the inflow that we have gauged over 112 years. Like I said, it's even steeper in Perth. It's even steeper than Sydney in Melbourne. Southern Australia is getting less. We've got a more limited supply in our catchments, yet population's going up. To the credit of the collective population, we're using less water per person. But the stress is there and we do manage that through water restrictions—as for both the millennium drought and Black Summer, and for the nineties drought. They can reduce it in the order of 20 per cent to 40 per cent, and that's named in that strategy. But, commercially, I really don't know beyond sending, say, a Sydney Water consultant to go and do a bit of a plan and look for toilets that are flushed too much or something, or changing the landscaping. One of my concerns—and it reminds me of an almond crop—is that, once you plant that, it's always going to need water. Data centres will degrade if they're not cooled. They have to be. CHAIR: Yes. They're not going to turn the air conditioners off on a 40-degree day, are they? Dr Wright : No. And it's not going to come from desalination. Sydney's got, for the size of our population, the smallest desalination. It's 15 per cent and it only goes to part of Sydney. We've only got one major abundant source of water and that's wastewater, so I keep coming back to that. CHAIR: Could I ask you—sorry, Senator Pocock—this: was it realistic when they said that these two upgraded wastewater plants, and two new ones coming online, would be ready to roll, with data centres being able to plug in within three to five years? Is that realistic? Dr Wright : I think that's a reasonable estimate. It could be a bit on the optimistic side, but I think it would be reasonable if it were a high priority. CHAIR: Yes. Back to you. Senator DAVID POCOCK: I might come to you, Dr Cumbo. In your view, is the federal government focusing adequately on the long-term benefits of these data centres? From where I sit, I hear all the talk about the initial investment and then I think, 'Okay, well, what happens after then?' Dr Cumbo : I think, in terms of value, it really depends. For instance, the AI training models—we're looking at the consultation papers—really shifted the language from co-locators and cloud storage to AI training now. That is looking at AI factories, which will likely not be in urban areas but raise other opportunities and challenges. In terms of the long-term value, again, I think it's playing on this optimistic idea of possibility, which we don't know yet. I think we're going to hear from people who have a lot more experience around AI models than me after this. But there's great potential with AI. I think it's more about being more discerning around what we want this AI for and the value of things like generative AI. What is the value that it's bringing our society, and is it worth the cost and trade-offs? As just a quick note on the regulation—sorry, Senator Ghosh, if you don't mind—I think, with regard to regulating, we need to stop focusing on the data centres. They've managed to hijack all of our values somehow. This is a place based problem: how can we think regionally about all the different uses of water, energy, land and space so all these animals and people can thrive and think about it in terms of these regional challenges. If you're living in Western Sydney, what is it like to live there? How are we catering for that with these new data centre clusters? At the moment, I think we're looking at it in terms of the data centres and how we can serve them, but it's hard—it's really hard—to coordinate and integrate. The other thing, in terms of long-term impacts, is regulating temporally—over time. These data centres have the potential to become obsolete, unless they're regulated to stay relevant. I've walked into three data centres in the last 12 months, and they've all been at least half empty. They've hit their energy caps. How do we make sure that these data centres are upgrading so they're more efficient and so they can benefit and work efficiently. That's a very costly process. I think more transparency around how they're going to do that over time, and potentially regulating for that, could be valuable and contribute to trust around what the industry is doing, because it's the great unknown at the moment. CHAIR: So you think we could be saddled with a whole bunch of stranded assets? Dr Cumbo : Potentially. Senator DAVID POCOCK: Just finally, Dr Wright, you mentioned the need for a national data centre inventory. On a separate committee looking at AI, I asked government agencies how many data centres we had in the country. They didn't know. They didn't know which were Australian owned. Yet they said that they probably would be a target in a military conflict. What are the sorts of things that we should know about data centres? Location would probably be a good start. What are the things that we should be looking at, and what sort of transparency should there be for the community, given some of the concerns about data centres and also the AI that they're powering? Dr Wright : I love having my seat at this table. I'm sure my colleagues here can add to this. This is what I was trying to find out. One source of information was these glossy brochures from the industry or from the state premiers seeking the investment, and the other was basically media. I'm trying to get the local communities' interaction or things like the utilities dealing with the communities of particularly the USA that have really copped it. But media articles have been really helpful for me for whatever brave people are left. I was looking for things like how many are planned, where, how big, how many are approved but not yet built, how many are operating and, just for my bent, how they are cooled. Air cooling is in the new one in Queensland, which I didn't think could be done on that scale, and I'm still not 100 per cent sure it can be. I was also looking at water use and how much is recycled; the power source—in many cases, they bring their own; carbon emissions; and also what sensitive sites they are near, community cultural, and how that is going. That's my laundry list. Senator DAVID POCOCK: Is there anything you want to add? Dr Cumbo : What was the question again, sorry? Senator DAVID POCOCK: What should a national inventory actually track? What do we need to know about data centres? What information is useful? Dr Cumbo : I think the usefulness of the data centre is something that I've been very interested in. CHAIR: What it's used for? Dr Cumbo : Yes. This critical industry—they fall under the banner of critical infrastructure. What are we supporting? And obviously there's privacy around data inside the data centres, but there's also the economic and governance arrangements of data centres. For instance, AirTrunk or Equinix or these data centre developers are landlords housing to mostly big tech, so there are layers of opacity in terms of who is using these data centres and what they're using them for. We don't want everyone knowing all our data. If we're talking about security risks, that obviously has to stay private. But I think we can definitely classify usefulness. We also need to think about redundancy. They have so much redundancy built into this system, with multiple data centres housing replicate data. If we think about the trade-offs with people and the environment, do they really need all of those things? How are they designed to mitigate—if one data centre goes down, often you won't even know. They have automation in place to just shift over to a different data centre. So, yes, it's designed around capitalising the value of optimisation, efficiency and speed. These are the things that drive the industry, and we just have to query the costs of that, particularly when we have critical issues. If we've got drought or fire or there's a disruption, how do we want to manage that? Dr Rey : From my side, I think: Who benefits? Who is going to benefit from this process? Has anybody discussed this at the public level? Who is going to benefit? CHAIR: I think that there's been quite a bit of talk of the analogy between data centres and particularly the large language model hosts, that this is data mining and that this is a new form of extraction. The question in my mind is that, in the mining industry, the community gets royalties. What does the community get out of hosting these facilities aside from just increased power bills and potentially less water? Dr Rey : And fire risk. CHAIR: And the shopping list goes on, yes. Dr Cumbo : We see that AirTrunk recently offered the community a number of different things for the Kurri Kurri data centre in Newcastle. There are a lot of community benefit agreements in place. But is it worth the cost? How was that benefit determined? Was this another top-down kind of thing that was decided by a few people around what the benefit could be? How is this long term? Is it a one-off beneficial scheme? This is how all the data centre developers operate. They often do have to do the community benefit agreements, but— Dr Rey : And is it sustainable? How long is this going to be sustainable for? If it's not sustainable, what are the costs, as you say, for that fact? Obviously, if it's sustainable for multiple generations and it's okay, then the cost becomes reasonable. But, if this is a 20-year or even a 30-year or 40-year situation and then it's discovered, 'Oh no, that was a mistake,' then the cost is going to be huge, based on that. We need to focus on sustainable futures and not short-term thinking. CHAIR: Thank you, all. We're going to have to move on to our final panel for today. Thank you for coming in and giving us your information. You've taken some questions on notice and we're very keen to hear what your responses will be when you can get them to us..