Examination of witnesses
Witnesses: Stephen Busette, Dr Alejandro Gallego-Schmid and Pablo John.
Q118 Chair: Welcome back, everybody, to this second panel session of our look into data centres. I will invite our panellists to introduce themselves, the organisations that they represent, and their particular area of interest and involvement with data centres, starting with yourself, Mr John.
Pablo John: Hi there. My name is Pablo John. I am Head of External Affairs at ADE. We are a non-profit trade body advocating for heat networks and other demand-side energy solutions such as flexibility, thermal storage and industrial decarbonisation. We have about 160 members across the heat network and demand space.
Chair: Fantastic, thank you very much.
Stephen Busette: I am Steve Busette. I work for Siemens. I am the lead for data centre development and markets within the UK and Ireland. Our speciality is really working with data centre partners and the wider circle with regards to introducing infrastructure and essential critical applications to ensure sustainability and efficiency throughout those networks. I am happy to be here to talk about this.
Chair: We are very happy to have you.
Dr Gallego-Schmid: I am Dr Alejandro Gallego-Schmid. I am a reader in circular economy and life cycle sustainability assessment at the University of Manchester. Part of my research is to quantify environmental impacts and apply circular economy strategies to different products, services and processes, and among them is data centres. I am also the head of sustainability of the School of Engineering and the head of the environmental sustainability working group at the University of Manchester.
Chair: Excellent. Thank you very much. I will start with my colleague Adrian Ramsay. I think you have an interest to declare ahead of your questions, and then the floor is yours.
Q119 Adrian Ramsay: Yes—thank you, Chair. I need to declare an interest as a vice-president of ADE, which is an unpaid role.
Thank you to the panel for being with us. I have some questions for Dr Gallego-Schmid and Mr Busette, please. First, what is your assessment of the current metrics and standards that are used to assess life cycle impact? I am thinking about things like materials use, e-waste, water usage and scope 3 emissions. Are those metrics adequate?
Dr Gallego-Schmid: We have to gain more transparency regarding that kind of data. As an example, the European Union is already asking medium data centres to provide data that we are not asking for in the UK. We are talking about the percentage of renewables associated to the electricity that they will consume, how much waste heat will be reused, and the power and water efficiency. These are basics.
We need to go beyond that; we need the kind of data that you have mentioned. We need to know not only how much water is consumed, but when that water is consumed. Obviously, more cooling is needed during the summer, when there is more demand for water. We need to know what kind of water is used. It is not the same to use drinkable water as using water that you have harvested directly from the rain or is coming from wastewater treatment plants.
We also need—this is at the heart of my research—to consider life-cycle aspects. We have quite a lot of focus on the operation part—we are talking quite a lot about electricity and water—but we have to start to talk about materials, about the embodied emissions and water that is used in producing the hardware, and about what happens with that hardware at the end of life, because we are generating more e-waste. Particularly in data centres that are associated with artificial intelligence, the life expectancy of the hardware is reduced compared with other types of data centres. This is the kind of data that we have started to consider, along with circular economy indicators: how much of that material is repurposed, reused or recycled? All this data is crucial to understand the sustainability of these data centres.
Stephen Busette: It is a great question. Data centres have changed over the years. I have been in this industry for 21 years and I have seen a lot of different things. I will give you a quick story. The first time I went to a data centre I was called by my boss as an engineer to go to site, and the first thing I saw when I went around the corner was two fire engines throwing cold water to keep chillers cold. That was almost the de rigueur association with that site. It sounds insane, but things have progressed quite wildly over that time and we have seen a lot of progression within the data centre industry.
To answer your question very succinctly, data centres now are looking at how they can reduce carbon. Carbon is an issue. Water use and energy use—these are all front and centre questions that are being asked of data centres. There is rigour about the supply chain, how materials are used in data centres, and whether they are recyclable. The life cycle of a data centre has been reduced wildly. There was a time when you could say a data centre would last for 10 to 20 years. Now we are talking about a five-year cycle. That goes with the reduction in scale, with the introduction of AI technologies and the density of chip matrices becoming much greater, and the energy being used for that. That is causing questions. For companies like Siemens, that question is distilled into, “What we can do to support and supply instrumentation and computation that will enable them to use it in a much more effective way?”
That is how we work with partners: to see how we can get consistency and some level of standardisation across the piece, and then how we can instil these models into workable models for the future. I think that the adaptability of data centre design is crucial to this. No longer are we seeing data centres as bricks-and-mortar houses; we are seeing a lot more modular designs coming through. There are reasons for that; it falls exactly in line with what you are talking about. If you build a data centre as a bricks-and-mortar design with a certain square footage, you will only get that square footage operating within that square footage. If you distil it into a modular design and modular pieces, you are able to control that, and to control its future and how it is scalable. That scalability is key to the answers to those questions.
Q120 Adrian Ramsay: Thank you both. You raised a number of points about both the immediate and the life-cycle impact, whether it is on energy, water or materials, which I am sure colleagues will delve into. The area I want to delve into a little more is water use, because I think water sustainability is a huge issue already facing the country—we have seen that in a number of reports in recent months—and this could make things a lot worse.
To illustrate what I am talking about, in Norfolk, a bit outside my constituency, there is an application for a data centre and there is a lot of concern from residents about the impact on water in a very water stressed area. East Anglia is famously very dry. We are certainly seeing the effects of that at the moment. Do you think that when a planning application for a data centre is considered, the sustainability of the water supply needs to be considered as a material consideration in the same way that impacts on flood risk or protected habitats would be? How should that be properly accounted for?
Dr Gallego-Schmid: Definitely. It is crucial. To make it clear, we have to differentiate the water that is consumed directly in the data centre from the water that is consumed indirectly. The direct part is the cooling, and that depends quite a lot on the localisation. Again, the period when you will have more heating in the data centre is the summer, when there will be much more in terms of other demand. It is crucial, particularly in areas that are water deprived, to consider the possibility of using systems for cooling that do not use water. There is a trade-off; normally, they consume more electricity, but this is important.
We also have to consider that a good part of the data centres that we have now in the country are located in the south-east—the area of London—which is an area that is water deprived during the summer. Many of the applications that we have in the pipeline that we have been talking about are also looking at establishing themselves in London. In those specific areas, water should definitely be part of the assessment of the sustainability of those data centres.
Stephen Busette: I can come at this from a little bit of a different perspective. Data centres and the technologies in data centres are changing wildly. We have 10x models and 80x models and all sorts, and what that means is that the density of the chip matrix is changing completely. A data centre was once in a server rack, 10 kW; you are now looking at 150 kW and upwards of that. Really, that is because the chip matrix that goes inside it is a lot more dense and has a lot more power and heat. That means that you will need different means of cooling down those chipsets.
One of the means of cooling them down is immersion technologies using dielectric fluid, which is an immersion tank that saves on water—it does not use water—and can be applied quite seamlessly within a modular design for data centres. We are seeing a lot more of that, because you would never be able to diffuse that heat by using conventional means like water. If you apply that application to this, you could be looking at a 30% or 40% saving on energy cost. That is an immediate impact with regards to energy, and we are seeing these designs being metered out within the industry.
What does that mean for us in terms of the future? It means that it probably will not be water based. The other thing that I can add to the argument is that we are looking at data centres that have closed loop systems. Instead of water going to the drain, water is cleaned, used and it goes around the same system. Once the system is filled, that is it. All these technologies are being installed right now.
The other crucial part of that is the logging of the data and the information. That is crucial to the operation and optimisation part of this. At Siemens we try to concentrate on that side of it to ensure that these systems are running in a purposeful and sustainable way. That is the real focus. What I am trying to say is that there is a technology piece. The things that we may have done in the past are changing rapidly. It would not surprise me if, within a couple of years, it is so modular that, almost like a mobile phone, you will take that piece of kit out, it will be repurposed for something else, and you will put another piece of kit in. That is the mentality that is going through the industry at the moment.
Q121 Adrian Ramsay: I hear what you say about different options and impacts of water, but of course the impact of energy is also substantial. Dr Gallego-Schmid, you are focused on life-cycle impacts. If we look at companies like Microsoft and Google in the tech sector, 97% of Microsoft’s emissions are scope 3, whereas Google has seen a significant increase in its emissions overall, despite scope 1 and scope 2 going down. Looking at the overall impact is clearly important, so what do you expect to be the impact of data centres in terms of overall energy demand, especially if we are not using water for cooling?
Stephen Busette: In this case I would think that the water demand would decrease. We are seeing a lot of that in the designs. We are very much seeing a focus on trying to find other means of cooling down systems. I will give you—
Q122 Adrian Ramsay: My question is about the implications for energy use. When the country is putting so much effort into trying to decarbonise electricity, could this bust our carbon budgets entirely?
Stephen Busette: I will answer that; I was getting around to that. You are absolutely right to voice this. Some of the question marks around this are about the energy use, completely. To give you a prime example, I am looking at projects at the moment where there are passive air systems that do not use water. They are able to operate in a very low energy density way, and the energy and heat are used to grow plants. It is an agri-data centre.
These are the technologies that are being used. There is a company called BlaqCat that is doing that at the moment and we are working very closely with it. If you need more evidence of this I can certainly provide it to you guys, because I think it is very good to look at these projects in a holistic way. If you look at that as a north star in the UK, with the technology that is available, the way that works is by being able to measure the carbon use, the energy use and the water use. It is very minimal water use. We are talking about less than 10%—
Chair: With respect, you have been asked about energy and we are particularly keen to get an answer on that. We are really excited by some of these new opportunities and we will try to find out about those, but can you respond to the question?
Stephen Busette: Yes, absolutely. You will see a vast reduction in energy use as result of these technologies.
Q123 Adrian Ramsay: Reduction compared to what?
Stephen Busette: Reduction compared to normal, conventional means, working with grid capacity, with regard to taking energy off the grid, as an example.
Dr Gallego-Schmid: If we are talking about the impact on electricity in general, the answer is that it depends. Take the example of the estimates of emissions for 2035. According to DSIT, data centres will account for 0.05% of the country’s total emissions. According to Carbon Brief, in an extreme scenario they could contribute up to 20% of emissions. We are talking about the same electricity capacity, so what is the difference? The problem is not the electricity itself, but how we produce that electricity.
One of the problems I see is that if we want to develop that important amount of extra electricity associated with hyperscale data centres, how are we going to cover the gap between data centres being developed and them being connected to the grid? In other countries that are a step ahead of us, such as the United States and Ireland, an important part of that gap is covered by natural gas, and that is where the emissions can change quite a lot. A data centre can be constructed in 12 to 18 months, but the connection to the grid can take years, so how are we going to cover that gap? That is the crucial part regarding emissions. We also have to be aware that the UK electricity grid, although quite low carbon, is not totally decarbonised. One third of the UK’s electricity still comes from fossil fuels—natural gas.
Q124 Adrian Ramsay: That is my concern: is it going to slow down that trajectory by adding a lot of extra demand that we have not planned for?
Dr Gallego-Schmid: We have to look at that gap, and it will depend on how renewable energy develops. It is also important how we plan things, and the role that data centres can have in the flexibility of the grid: how they can absorb energy when we are producing extra energy from renewables and what we can do when they are consuming at the highest levels at the same time as we are using more renewables. That is the key part that will determine how this affects the final emissions.
Stephen Busette: Added to that is the deployment strategy. Being able to deploy these technologies in the right place and at the right time is imperative. We often talk about the infrastructure that exists versus the infrastructure that would enable these technologies to work. The infrastructure that would enable the technologies does not necessarily exist in its standard form today. Perhaps the standards need to creep up a little towards that.
Chair: Thank you. Before I call Chris Hinchliff, my colleague Martin Rhodes needs to declare an interest.
Martin Rhodes: I just refer to my entry in the register of interests, in that I am a vice-president of the Association for Decentralised Energy.
Q125 Chris Hinchliff: I have two quick questions to take us back to the discussion on water. I accept what has already been said about the technological innovations that have been made there, but it is worth setting out the context of people’s concerns about this country heading towards water bankruptcy and a deficit of 5 billion litres a day. Water companies have written to me highlighting data centres requesting 35 litres of water a second, equivalent to the peak demand of 3,500 homes. We have seen in Chile—I will try to get this right—that the expansion of major data centre hubs alongside drought conditions has pretty much devastated the Quilicura wetlands there. We need to make sure that similar things do not happen to habitats in this country. Do we need to mandate more water-efficient technologies to allow this expansion to go ahead?
Stephen Busette: Gosh, yes. It would be crazy for me to sit in front of you and say, “No, of course not.” To give a clear answer to everyone, these structures exist because of what we have right now. We are facing great challenges around the fantastic idea of AI, AI growth and so on. It sounds wonderful; technology is brilliant, it adds to the economy, it builds jobs and all these things. However, there are those questions being asked. If we look further afield to other countries where these things are happening—God forbid it happens in the UK—we are in a position now to learn from those experiences and use our wits to create models that work, so that we can deploy these things effectively.
Q126 Chris Hinchliff: Thank you. That was a very useful and clear answer. Picking up on what was said earlier, where data centres do have water demand, that often peaks at the same time as water demand generally is peaking, when it is hottest. In the probably inevitable circumstance where water runs short in a particular area and water companies are running out of supply to meet all the demands they face, does the panel think there should be a situation where we agree that the water to data centres is reduced before we start turning off the taps to the homes of our constituents?
Dr Gallego-Schmid: I hope that if we do things correctly and plan things correctly we are not going to be in that situation; that should be the aim. Obviously, it is not my decision whether we put people before data centres. My personal opinion is yes, but it is not for me to take that decision. Something we have to be aware of is that people are very sensible regarding water. Is this going to mean that the water is of lower quality for my kid, for example, and that am I not going to be able to shower? That is the key issue, and we have to address very well for local communities the potential implications of the development of data centres. If we are at risk of the kind of extreme case that you mention in Chile, and that kind of effect on biodiversity, then we have to put a limit on it, and maybe that is not the place to put a data centre.
Chris Hinchliff: Does anyone else want to comment?
Stephen Busette: I think that is very clear. Those examples are things we need to learn from. My personal position is that I would hate to see that. From a business point of view, and certainly from a wider-scope point of view, do we therefore place our leanings into data centres as a foundational growth engine within the UK at risk by just saying we are going to put a moratorium on data centres? That is almost a non-fanciful question, to a certain degree.
Q127 Chris Hinchliff: I suppose I was not asking whether we should put a moratorium on data centres, but whether there ought to be regulations in place to say that in periods of extreme water stress—and this is already happening; the taps in my constituency went off during the brief heatwave in May—we prioritise the public ahead of business use, but data centre use in particular, because of its extremes.
Stephen Busette: I think a lot of that comes with planning and foresight, and the deployment strategies and so on. If we get that right, then that should not even be a question. If we look at closed-loop environments and things like that when creating these data centres, then that should not be a problem, because that would form part of the strategic basis. That is the kind of thought pattern I have around this.
Chris Hinchliff: So you are optimistic that we can avoid it. Mr John, do you want to add anything?
Pablo John: I am not as much of a water expert as my colleagues here, but I would add from an energy perspective that if you want to use the heat in a heat network, you will want to use a closed-loop or a direct system. The systems that reduce water use are also the best for heat reuse, so they have a double-whammy benefit in that way. We would very strongly support it.
Q128 Jonathan Davies: I am mindful of time; I think there may be something on at 5 o’clock that has some people’s interest.
Stephen Busette: I wonder what that is!
Jonathan Davies: I have a very dry meeting at 5 o’clock, so I will be missing the first—anyway. Mr Busette, you have touched on some of the technologies that can be used to reduce the environmental impact of data centres, whether that is in terms of the energy they use, the water they use or the heat they generate. Setting heat networks aside, because I think another colleague will come to that, are there any technologies you have not referenced in your answers so far that can reduce the environmental impact of data centres? I know you have talked about immersion and closed loop, but is there anything else you think we should know about?
Stephen Busette: Absolutely. In data centre design, and certainly data centre build, the planning phase is very important. In the beginning phase, by using technologies like digital twinning before you go into the actual scale of understanding and deploying a data centre, you can do a lot of that pre-process work ahead of putting the data centre in situ. That is something that is readily available in the marketplace. It allows for a very clear understanding of what the scenarios could be. For instance, if in your particular environment you have a certain amount of water, you can put that into a digital twin model, and that will give you the answer as to whether you should put a data centre there. These are models and things that already exist.
There are tools for power optimisation, especially with regard to building management systems, which can easily work out the most efficient course of action regarding how power is generated, what it is used for, what should be turned on or off, and so on. All those things are readily available. The biggest part of this is that these optimisation models, the equipment and the AI compute involved in this process are there to be had; it is whether or not you strategically have a measure to say, “We’re going to deploy this at the beginning stages, in the planning phases.” That enables you to say, “Okay, I will pick liquid cooling here, I will have a battery-powered system over there, I will use renewables there,” and to pick and choose in tandem where these technologies fit. I think that is the way to go about it. If I were given the quest—I am not a data centre builder by any means; we are a provider of technologies into that sector—that would be the way of going about it, in my estimation.
Q129 Jonathan Davies: Data centres obviously have huge opportunities for the economy, to make public services more effective and to help the environment through modelling. We have also heard that, largely perhaps because of behind-the-meter sources of energy generation, they could account for 8% of UK emissions by 2035. Do you think these technologies will mean we will not get anywhere near that, or do you think that 8% is a conservative estimate?
Stephen Busette: I really can’t comment on that at this stage. I have certainly seen some of these arguments, but I would say that by deploying these types of systems you can reduce these factors, though it really would take a concerted effort around the deployment of these technologies within the sector in order to—
Q130 Jonathan Davies: Do you see much commitment from the sector? Is this embedded in their thinking?
Stephen Busette: It is in every conversation we are having at this moment in time, if I am honest with you. Sustainability and the environment are always ahead of the task. They are very aware of the conversations happening within Government and very aware of the viewpoint—the optics. Every single person I have spoken to over the last five years or so has been embedded in this conversation.
Q131 Jonathan Davies: What are the barriers to scaling up those new and emerging technologies to mitigate the environmental impact?
Stephen Busette: Some of the barriers, it seems, are really based around the conversation pieces like the ones we are having today. There is a real thirst for forging forward, but it is whether or not there is a vital appetite to really commit to the technologies available to produce the results. It does take planning, it does take real guts to get to those scenarios and answers, but if we do that from these stages we are in a good place to deploy these technologies.
Q132 Jonathan Davies: Do you think firmer regulation and more clarity around regulation might help? At the moment, there is DSIT, but there is not necessarily a regulator for data centres.
Stephen Busette: I cannot speak for the data centres themselves. Regulation is something we have not necessarily got a handle on, and I guess that is a question for you guys on the Committee to answer after this. However, from our position and what we can see and do, we can certainly help with the conversations around this and with the instigation of the technologies to deploy in this space.
Q133 Jonathan Davies: Thank you. Dr Gallego-Schmid, do you think our approach to managing the environmental impacts of data centres is too heavily reliant on these emerging technologies that we may not fully understand or appreciate yet? Are we overstating the benefits of some of these technologies?
Dr Gallego-Schmid: I am not an expert in those technologies, but I think what we have to try to do—and this is related to what I mentioned about analysing whether those technologies are useful—is to gather more data and have standardisation. That is one part I think is missing, which would allow us to quantify data centres from the sustainability perspective. That is the part that I think is missing. For example, it is not yet approved, but the European Union is starting to develop a label similar to what is applied to electronic devices—B, C, D, E—based on power efficiency and water efficiency. It can be really helpful to see, if you are applying those technologies, whether you are achieving the higher value. That is really important for local communities, but also for the public sector, because public procurement is very important. Having those standards will help include sustainability criteria in public procurement.
Q134 Jonathan Davies: If I might ask you one more question—and if any of the other panellists have views on this I would be interested to hear them—I think I am right in saying that instructing an AI system to find some quite basic information uses 10 times more energy than a regular Google search. In other aspects of life we talk about modal shift. Are we using AI for things we do not really need to use it for? What can we do to incentivise the consumer not to use AI when there is a more straightforward, cheaper, less energy-intensive solution? Do you think there is a perverse incentive on behalf of the people behind AI to push it in a way that means that we are using it when we could just look in a book or google it?
Dr Gallego-Schmid: I am so happy that you asked that question, because it is one of the points I wanted to raise here. We focus quite a lot on what we might call the production side, but not so much on the demand side. Everybody understands the benefits of artificial intelligence—it can help tackle cancer or climate change in the future—but that is not all the demand we have on artificial intelligence. I read the other day that ChatGPT receives 2.5 billion requests every day, and I do not think all those requests are related to these valuable and honourable goals.
I am a person who is pro-freedom—I am not in favour of restricting the use of artificial intelligence—but we should inform the public so that they can make good decisions: telling people, for instance, the environmental implication of using artificial intelligence to generate an image or a video. Companies can also do more in that sense, in that sometimes it seems artificial intelligence is imposed on people. If I do a web search, in many cases I already receive an artificial intelligence response that I did not ask for. Imagine multiplying that for everyone. I think that is what we should tackle, because that will reduce the demand and reduce the impacts associated with data centres, while we still get all the good things from artificial intelligence.
Q135 Barry Gardiner: Mr John, you must have been feeling like Billy No-Mates down there, so let’s turn to you. I think you were here for the first panel, so you will have heard the discussion about the increase in a city’s heat, of between 2° and 9°, that my colleague Julia Buckley was talking about. Why are we not capturing that?
Pablo John: I was actually quite glad not to have had your eye on me before, but there are a few different barriers. Fundamentally, it is political. As you know, the Government have a Clean Power 2030 target. That has been really helpful for our colleagues in the clean power world, because they know they can invest in clean power infrastructure and there will be political support for that in the short to medium term. There is no statutory target for clean heat. When you hear the Secretary of State talk, it is about things like nuclear and offshore wind—all very important things—but we very rarely hear them talk about heat.
Q136 Barry Gardiner: How can we take this thing that we are all talking about as if it is a problem with data centres and turn it into a resource and an advantage through district heat networks and so on?
Pablo John: The most impactful thing Westminster could do in the short term to promote heat networks is around electricity levies. You might remember that in the Budget the Chancellor removed electricity levies from energy bills, taking £150 off bills. That was very welcome, but it only applies to domestic households. If you are on a heat network, you still pay a £150 levy on your energy bill, even though your neighbour on a gas boiler will not. That makes it very hard to close the cost gap between low-carbon heat—including from heat networks—and gas. If those levies were extended to include non-domestic households and heat networks, they would be much more competitive overnight, which means they would be a lot more investable. It would also mean that things like data centres would pay less if they used clean electricity, which would bring the price of that down. That is the best short-term thing this place could do. There are also things like spatial planning and heat network zoning—
Q137 Barry Gardiner: I want to be clear that I understand you, because it seems to me that this is framing up into a recommendation the Committee might make. At the moment, you have a problem. The problem is heat and you need to cool things down, so you are spending money and resources dealing with that problem, that heat. Why would it not be in the commercial interest of a data centre to say, “Hey guys, we’ve got a resource here. Why don’t we sell it to somebody? Why don’t we get a district heat network that can take our waste and use it as a resource?”
Pablo John: Because there fundamentally aren’t the heat networks. In the UK, 3% of our heat demand is met by heat networks. We should have been building this stuff 50 years ago. Unfortunately, we have not.
Q138 Barry Gardiner: I always like the advice that says, “Well, I wouldn’t be starting off from here,” but it doesn’t help us go forward. What can the Government do to actually incentivise you guys—I am saying “you guys” but I mean the guys in the first panel, the data centre guys—to be co-developers, because that would solve a problem for them?
Pablo John: Absolutely. There is a policy the Government are working on now called heat network zoning. Colleagues might be familiar with that. The big problem you have with a heat network is that if you are a consumer, you want heat—that is very fair—if you are a heat provider, like a data centre, you want someone to pay to take that heat, and if you are a heat network developer, you want both consumers and heat suppliers, but no one wants to go first. You do not want a heat network with no heat, and you do not want a heat network with no customers. The idea of heat network zoning is to end that stand-off by forcing everyone to move together. Legislation is going through Parliament on this now.
Our big concern with heat network zoning is that, as part of it, local authorities will have to appoint heat network co-ordinators, someone whose job is to say, “There is a data centre here and heat demand here; you guys have to link up”. That is a very onerous technical role—you need to be one part urban planner, one part engineer, one part communications professional—and there is no ringfenced funding for local authorities to do that, so you are basically asking local authorities to become mini-regulators of heat networks with no extra funding. I think empowering local authorities and ringfencing funding for them to zone for heat networks so local authorities could say, “All of you move together,” could make a huge difference.
Q139 Barry Gardiner: So that is a recommendation that you would like to see in this Committee’s report.
Pablo John: Correct.
Q140 Barry Gardiner: Thank you. Should heat reuse for data centres be mandatory?
Pablo John: I would say yes, but I would have a lot of caveats.
Barry Gardiner: Good—give me the caveats.
Pablo John: If you are building in a heat network zone—if you are building in inner-city Leeds or west London—absolutely you should be made to connect. If you are building a data centre up in, say, northern Scotland to try to access curtailed energy and use that demand flexibility, where there is not the demand, you should not.
There needs to be a bit of a carrot and a stick here. What are the two things data centres want from Government? No. 1 is grid connections; No. 2 is planning permission. Data centres are now national critical infrastructure, which means the Secretary of State will make a lot of these decisions. When making these decisions, the Secretary of State should take into account things like waste heat reuse, flexibility, co-location and thermal storage. If a data centre says, “We will share our heat with the community. We will take excess energy from the grid to prevent curtailment,” that should count in its favour. It should be bumped up the planning permission and connections queue.
Q141 Barry Gardiner: That is a lovely suggestion. I think that, again, is a recommendation this Committee might make: to have a hierarchy for getting on with it—
Pablo John: Yes, and the same for grid connections for Ofgem.
Barry Gardiner: So if you are joined up with the district heat network, you go to the top of the queue.
Pablo John: Absolutely, both for planning permission and for grid connections.
Barry Gardiner: Well, maybe not the top of the queue—[Interruption.] A points system, yes.
Pablo John: Absolutely. It should be considered and should work in your favour.
Q142 Barry Gardiner: Now, I have a problem, haven’t I? We have just seen the weather forecast that the Met Office did for 2056. You may remember that last Monday the Met Office held a big event where it said the peak summer temperature in 2056 is going to be 45 degrees, and the lowest temperature anywhere in the UK is going to be 30 degrees. How are heat networks going to operate then? What do you do when there is too much heat everywhere else, it is peak summer, and nobody wants to use that heat for a heat network? What do we then do with it? There must be lots of other uses for that heat. What could they be?
Pablo John: I think there are two big uses. The first is thermal storage. For those who do not know, thermal storage is essentially a big kettle. There is a big vat of hot water, and when there is a lot of heat and the heat network does not need to discharge heat, you instead charge the thermal store at times when electricity is cheap. Then, when it comes to winter, you can discharge that energy much more cheaply for your customers.
The other place the heat can go is for industrial use, either through a thermal store or through co-location. There are rural data centres, as I mentioned earlier, that could co-locate with a distillery, so the excess heat from the data centre could support the distillery.
Barry Gardiner: You are speaking my language, Mr John. That sounds very good.
Pablo John: One of the world’s first net zero distilleries is up in Annandale, in Scotland, and it uses a thermal store. It does not use data centre heat, but it takes excess energy from the grid on a windy night in Scotland and charges the store, which means they get cheap electricity, they get to decarbonise, and we get delicious whisky.
Barry Gardiner: As a good Scotsman, I don’t give a fig about what is happening at 5 o’clock this afternoon, but—[Laughter.]
Adrian Ramsay: You’ve got another 20 pages, Barry?
Barry Gardiner: But I think, if I don’t want to incur the wrath of the Chair, I had better leave it there.
Chair: Very wise. Sojan Joseph.
Sojan Joseph: Thank you, Chair. I will be very brief.
Chair: Take as much of the next six minutes as you need. [Laughter.]
Q143 Sojan Joseph: Thank you. Mr Busette, I heard you talking about how efficient the technology has become. We have heard that from various other witnesses, too. The Committee has also heard, and some of my colleagues have already discussed, how much impact this is having on the environment, energy, water, heat and all that. In my constituency in particular, we already have a hosepipe ban from South East Water, and people were not getting water when we had two days of high temperatures a couple of weeks ago. It is a real issue for the public, who are not talking about the benefits of data centres; they are worried about not getting water or electricity. That is their concern. How should the UK prioritise where, and for what purpose, data centre capacity is developed?
Stephen Busette: There are a lot of questions on this, and you are absolutely right. You almost want a map of the country of where the grid capacity, the water capacity and all these levers are, because at this moment in time I would not say it is a scattergun approach, but it is certainly like, “There is a bit of land there. It looks really good. It is quite near to power. Let’s build a data centre.” I personally think there should be much more planning and much more involvement and engagement by local councils with regard to this.
As an example, I have seen a project in Warrington where they repurposed an old factory that has a hydrogen intake of energy as well as electrical grid capacity to build a data centre. That ticks so many boxes. They are doing a closed-loop system and using liquid cooling technology in that space. They are not going to use water; they have a hydrogen connection and they have grid capacity. These are the kind of favourable places that we should be looking at. There are so many factories around the country and so many built environments and defunct sites that we no longer use. We should be looking at capacity in that way, so that places like your constituency do not necessarily have to have the added interest of a data centre adding more stresses on to the system.
We should look at it in a holistic way with partners and with people who have a vested interest in the technologies that go into these spaces. It needs to be a joined-up conversation about location, location, location, and we should really look in some detail at deployment and strategies around deployment.
Q144 Sojan Joseph: So at the moment there is no plan or map of locations in the UK? Dungeness near Folkestone, for example, is getting decommissioned. It is well connected to the grid, sitting next to the seaside. Is that a location that the Government should be looking to build?
Dr Gallego-Schmid: I think that is the purpose of the artificial intelligence growth zones, and the idea is positive. What I might criticise, in my personal opinion, is that the criteria for the growth zones are focused more on techno-economic aspects and not that much on sustainability. It seems like sustainability is a secondary aspect.
Related to that and to what you were mentioning, we also have to consider more the opinion of local communities. We have cases here in the UK—for example, related to fracking—that show what happens when you do not consider local communities. Obviously, everybody understands the benefits—I mentioned them before—of artificial intelligence, but this is going to create impacts. Local communities want to know what is there for them in order to compensate for those impacts that they are going to be suffering. We have talked about general impact, but not so much about specific local impacts. These facilities produce noise, air pollution and increased heat, and, sorry, but they are not really nice buildings either, and people can be against that. We have to start to address these local impacts, too, and involve the opinion of local communities more. In the States, where they are a step ahead of the situation here, there are already surveys that say that 71% of Americans are against having a data centre in their local community. That is something that we have to start to address in the UK and give reassurance to local communities about the potential benefits of having those data centres nearby.
Q145 Sojan Joseph: The counter-argument would be the benefits of data centres go back to the people as well. There is a huge demand for data centres. That leads nicely to my next question. To what extent should the policy focus on managing demand rather than just building as many data centres as we want?
Stephen Busette: If that is the lead-off question, I think that we should really look at where they are, what the demand is and what the plan is for UK plc, as it were. If we look at this in terms of the growth plans for data centres, the UK capacity space is probably double the size of its nearest competitor. The reason that demand exists is that the UK is seen as a favourable spot for this technology to grow and flourish, which, as you say, will aid the local community if we get this right, but we need to get the strategy around deployment right. It is essential to the goals of the data centres themselves, because they want to grow and to be able to bring that capacity to the UK, but it is essential for us, too, to look at the future and the benefits that we will have as a result of doing it properly.
The technology only comes into the space when we have done that, to be fair. You cannot put great technology in bad spaces. You want to put great technology in great spaces that have been devised with the purpose of doing it in the right way. If we do it in that way, we will be able to gain the benefits of that. We spoke about energy reuse and heat reuse. If you put it in a place where we actually will get the benefits of that, it will grow and flourish. If we get those partners talking in the right way, being really interested and having some truck in this, some stakeholders in this, then we will be in a position to be a beacon of light for the rest of the industry in this way of looking at things.
Your question is the crux of the matter, to be quite honest, in my opinion, because I think that we are at a pivotal point in the development of data centres where there needs to be some harnessing of the energies, not just from a business point of view but from a real political standpoint, around how we are going to make this grow properly. If I look at other countries that are doing these things, like Germany, where they are giving off 13 TW of energy per year, they are giving that back to the public. Why aren’t we doing things like this? You almost look across the fence and go, “Oh.” I want to feel as if we are doing all these things.
Pablo John: This feeds into one of the most important energy policy decisions being made now, which I imagine many Members in this room might not be aware of, which is RESP: regional energy strategic planning. NESO, the National Energy System Operator, is dividing the UK into various strategic regions being managed by a RESP board. The job of the regional energy strategic planners will be to identify where the demand is coming from and how we can get the supply, the wires, the cables and the energy from A to B.
As we have talked about, data centres are going to be one of the biggest pushers of demand in a generation. How many times do you think data centres are mentioned in the RESP consultation methodology? Five times in a document that is over 100 pages long. Our big concern with RESPs as they are being written now is that they are very much supply focused. They are very much about how we get energy from A to B—how we build the wires and the pylons to get energy from A to B. It talks about power plants as assets; it talks about demand as sites. Demand can be just as much an asset to the grid and to regional energy planning as supply can be, whether it is through heat reuse, demand side flexibility or thermal storage, but regional energy plans, as they are being developed now, are not taking that into account.
Another recommendation for the Committee would be that regional energy plans need to have a much stronger role for all demand, not just data centres but industrial sites, thermal storage and battery storage. It cannot be just supply-side: “How do we get the electrons from here to here?” That is a very old-fashioned way of looking at our energy system.
Dr Gallego-Schmid: I fully agree with the comment from Mr Busette, but I also think we have not only to see where things have been done correctly, but places where things have gone wrong. In Ireland, 21% of the electricity consumption of the country is used for data centres; 48% of the consumption of electricity in Dublin goes to data centres. Accordingly, in Dublin, there is a moratorium. There is a one-year moratorium in the state of New York, too. We have to analyse what decisions we are going to take, and this is the position that the UK is in now. That is, I guess, the reason that we are here.
It is not that there is a solution that fits everyone. There is a lot of talk about how many of the data centres that are in the pipeline are speculative, but Ofgem has already identified that half of them are credible because they already have finance commitments, and they represent 20 GW of energy. Just to put it in context, that is the amount of electricity consumed by 40 million people. We have to analyse, of those data centres that are in the pipeline and are credible, which ones we want to go ahead and to be on top of the pipeline. As was mentioned before, maybe sustainability can be a criterion. That is the crucial decision.
I do not think that we have to take this in an abstract way. We have to analyse each case very carefully because we are talking about massive engineering buildings. I think we all think of data centres as like the server room that we have seen in businesses; no, these data centres we are talking about are massive engineering buildings. In each case, decisions have to be made and we have to analyse what is the best location and what are the best technologies available.
Chair: Thank you very much, Mr John, Mr Busette and Dr Gallego-Schmid, for your evidence and some excellent recommendations. We will bring this session to a close.