Luisa Cardani, Steven Campbell-Ferguson and John Abraham.

Q110 The Chair: Good morning and welcome to the Lords Committee on Environment and Climate Change. Today we will be taking evidence from two panels of witnesses for our inquiry into drought preparedness, focusing first on water-intensive industries—I thank our three panellists for being with us today—and secondly on local drought response.

Before we start, I remind everyone that the session will be webcast live on Parliament TV and that a transcript will be taken and made public. Witnesses will have an opportunity to review the transcript and, if necessary, make very minor amendments. Members are reminded that any relevant interests should be declared the first time they speak. I will take the opportunity here to say that I am a director of Peers for the Planet and that I jointly own a borehole in our home in Oxfordshire.

Before starting with the first question, can I thank our witnesses for taking the time to be with us today and ask them each to briefly introduce themselves?

Steven Campbell-Ferguson: Good morning. I am the global product design lead for NTT Global Data Centers. We are a co-location data centre developer and operator. We have facilities in the UK in Dagenham, Hemel Hempstead and Slough, but we also operate facilities all over the world.

I have been involved in the design and construction of data centres for at least 20 years. In a previous life, I was a mechanical engineer. I have been designing cooling systems and water treatment systems for cooling systems for almost 35 years now in various other industries. Probably also relevant is that I am the water working group chair for the Climate Neutral Data Centre Pact, which is an industry grouping dealing mainly with the European Parliament and regulation around advising on best practice standards for data centre sustainability metrics. One of those metrics is a water target. That probably summarises who I am and why I am here.

Luisa Cardani: Good morning. I am the head of the data centre programme at techUK, which is the trade association for tech. We cover everything from clicks to chips, as they say, and I am here representing our digital infrastructure members, in particular data centres. TechUK also holds the umbrella climate change agreement for the data centre sector. We also have in-house our data centre council for the UK, which is around 20 industry leaders setting strategic priorities and reviewing things such as sustainability metrics.

John Abraham: Good morning. I am the chief operating officer for Veolia. I sit on the northern Europe zone board. I represent all of our industrial, water and energy operations across the UK, Ireland and the Nordics. I am a fellow of the Institution of Civil Engineers. I am an honorary fellow of CIWEM. I have been in the water industry for 35 years, with lots of miles on the clock.

I am delighted to be here with you today to use some of the extensive knowledge that Veolia has globally. We are the largest water services business in the world. We provide 111 million people with drinking water, and nearly 100 million people with wastewater treatment services. We work in many very drought-stricken areas throughout the world. We have a technology group. We have asset management, operation and maintenance teams all around the world. I am delighted to be here today.

Q111 The Chair: Thank you very much indeed. I will start off with the first question. I want to understand the key risks of drought for your sector, or the sectors that you engage with, and how these risks are being considered and mitigated.

Steven Campbell-Ferguson: If we look at drought risk, given that we need water for our operation, we probably do consider being off for a duration as a risk. In the UK, we would see the likelihood as very low, albeit the impact could be very high on sites that use evaporative cooling systems, which not all sites do. In a way, it is a risk, but it is seen as low.

The risk we do see in water and water supply impact is that of the pipe being dug up, giving us a short-term loss of supply. Generally, the industry will design to mitigate that by storing 24 hours of capacity on-site. The typical design standard is to have 24 hours of capacity. If there is an outage for longer than 36 or 48 hours, sites will begin to be impacted in terms of their ability to operate. The exact impact will depend on the type of cooling system that is deployed on that site.

The Chair: For a cooling system that is dependent on water and on the public water supply, what are the risks in a drought situation?

Steven Campbell-Ferguson: Again, in a drought situation, if we ended up with that supply being interrupted for greater than 36 or 48 hours, and it is an open cooling tower system, the data centre would need to shut down, because it would be unable to cool itself.

The Chair: What I am trying to get at is how you assess that risk. Is that a live and growing risk to you or are you confident that water will always be supplied to the sectors that are dependent on it?

Steven Campbell-Ferguson: To be blunt, the industry is probably confident that water would always be supplied. The very fact that we are here today maybe changes that view and so, in some respects, we will be quite interested to see the report that you produce, being some data that would quantify whether that risk is higher than we realise.

The Chair: Do you have conversations with any data centres that are dependent on the public water supply to say, “In future, it could be that we are not able to receive that water. Should we be looking to become a little more self-sufficient?”

Steven Campbell-Ferguson: We have at least two sites that are dependent. If that risk is higher, we would have to look at further mitigation technology.

The Chair: We can investigate that further down the line.

Luisa Cardani: Further to what Steven has already said, some of the risks tend to stem from drought combined with high temperatures, for example during heatwaves. Heatwaves tend to put stress on the infrastructure. It becomes more energy or water-intensive to cool the chips, so that is one risk. Of course, it can be mitigated in addressing or estimating the level of drought or heatwaves. If you have knowledge of this, you may be able to make adjustments, especially at the design stage, such as a trade-off between more energy-intensive cooling technology and technology that might use water.

That trade-off, of course, then has the impact that you are using more energy. If you are using more energy-intensive cooling systems, therefore saving on water, that puts more strain on the grid. It has an environmental impact as well, because it depends entirely on the mix of power that you get from the grid, which is not always 100% renewable power.

Furthermore, if you plan to build a data centre in a drought-prone area, there might be risks around reputation, because the planning authority might be a bit more cautious of the investment. That might bring more scrutiny. There should be more scrutiny, but this can delay the entire planning process, which delays investment.

Finally, there is always a compounded risk, since the sector has a tendency to cluster in certain areas. When there is drought combined with a clustering effect, all the risks can be amplified for the sector.

John Abraham: There is significant risk going forward. As a solution provider, we focus on the solutions that we can provide to alleviate those risks. Specifically, across the industrial sector, there are many solutions. It is just the economics of the solutions that they would need to adopt, and I am sure that that will come up later in the discussion.

Some of the real risks that we see, perhaps on a smaller scale, are a bit of a hybrid, in that you would not necessarily consider them industrial. We do a lot of water provision for the Ministry of Defence across the Wiltshire Army‑basing projects for one of the largest MOD PFIs, called Aspire Defence. For instance, we have boreholes that we are having to take out of service by 2032, because the aquifer is in such a state from the impact of drought. Therefore, we are working with the Ministry of Defence and the DIO in looking at alternative water supplies, water reuse, grey water reuse and lots of different solutions going forward. There is a huge risk, but we are used to managing these risks. It is an economic problem, not necessarily a technology problem.

Q112 Lord Rooker: Good morning and thank you for coming. I want to explore how much of a consideration water availability is when data centre operators select site locations. It might be more than one operator and more than one location. I will take the example of the one that we know about, which is planned to be the first AI growth centre, at Culham in Oxfordshire. Part of our briefing tells us that it will have no or negligible effects on Oxfordshire’s water supply, but I also read that it is seven miles away from what will be a new reservoir at Abingdon. That reservoir is intended to supply the Thames Valley, London and Hampshire. This is a very stressed area. It is in the Home Counties.

Britain is a very unequal country, so we are going to end up with a lot of data centres in the south-east, basically. How can you prevent one data centre being affected by another? We are going to get planning problems and they are going to start complaining that they cannot build their data centres, because the planners will not allow them, because there is already one in that location. How are they going to take that into account?

Luisa Cardani: I am very happy to start and then perhaps Steven can supplement. When it comes to site selection considerations, the main things are power, fibre and the ping latency that the main or anchor customer is willing to withstand. Especially since there has been a lot more data available about water-stressed regions and drought, water is increasingly being considered as well within that triangle.

However, it is worth noting that power remains the main thing and the main consideration. When it comes to water, that then brings design considerations. Operators might decide, based on the cost of power and the availability of water, or how stressed a region is, to make specific design choices, such as using waterless cooling technology, whether that is a closed-loop cooling system or a direct-to-chip cooling system. I cannot comment on specific examples. However, in theory, if you use a waterless cooling system, you will not be using any water aside from, perhaps, in the operator’s office and facilities.

In a way, in theory, if an area is stressed, planning officials should have conversations with new investors in the region about whether they would be considering more energy-intensive cooling systems rather than water-based systems. That is normally the way that the sector would look at this problem. It is worth noting that, of course, it is entirely a commercial choice at the end of the day, based, as I said, on things such as power availability. If you decide to use a more energy-intensive cooling system, you will be using more power. Depending on the power availability in the region, that might also be something that you need to consider.

Lord Rooker: Since you mentioned it, is there an example of a non-water-cooled operation where it is done in another way?

Steven Campbell-Ferguson: I will give an example, but, just to answer the initial question of what we do as a developer, we will look at a site and talk to the power company about power connection and what is involved, so capital costs to get there and when it can be done. That is a key driver. If we want to use water, we will talk to the water company and say, “Can we have a supply? Can we have it this big? Is it an issue?” Generally, if I went and asked today, the answer that I would get is, “No, you can’t have a connection”.

That makes the decision for us that we will then have a dry cooling system. It is a slightly more energy-intensive system, but there are lots of solutions. We do not need to use an evaporative system. Some people in the industry choose to use an evaporative system when the water is available, because it saves electricity and, overall, it makes a better economic business case.

In today’s environment, to give an example, which I think is in the public domain, we have recently acquired a site near St Albans.

Lord Rooker: I want to be specific about this. I realised that you said that these are commercial operations and, therefore, as a trade body, you are not going to give us individual details, but is there a specific example of this technology? If so, where is it?

Steven Campbell-Ferguson: There are plenty. More than 50% of data centres in the UK are operating non-evaporative cooling systems. Over the last 20 years, systems have typically been designed probably about 50:50. Twenty years ago, everything was dry because it was the cheapest way. The original dotcom boom 25 years ago was dry systems, because it is lowest capex and people, frankly, did not care about energy efficiency. The industry then got very interested in energy efficiency and water-using systems. Now, the pendulum has swung back, in that we cannot get the water, so we are going back to dry systems. There is a bit of history that water-using sites will tend to be of a specific age.

Lord Rooker: John, can we hear from Veolia?

The Chair: We are at 10.16 am.

Lord Rooker: I know that we are, but we should hear from all three witnesses, if possible.

John Abraham: I will be very brief. The economics of the cost of water versus the cost of power is at play here. It is 10 to 15 times the cost of energy versus water. When you have a closed loop, your energy costs go up, which is significantly more. If you can get water from the local network, you will; if you cannot, you will not. That has been explained very clearly. The technology exists for all the closed-loop techniques.

The other advantage of a closed-loop technique is that, if there is a district heating network or other off-takes for the heat, there can be a huge environmental and sustainability benefit from taking that heat from the closed loop that generates hot water and providing heat to homes in a very sustainable way. There are added benefits from a closed loop as well, which we are all working on and I would like to table.

Q113 Lord Krebs: My question builds on what we have just been discussing. It is about water-use efficiency. What we are told is that new data centres will be designed to meet a maximum water-use efficiency of 0.4 litres per kilowatt hour in areas with water stress. I wondered whether you could expand on this.

I want to contextualise my question with two comments. First, I have heard it said that, every time you do a Google search, you use a bottle of water. Is that true? The second thing that I have heard said, which echoes what Luisa told us, is that, if it is true, most of that bottle of water is used in generating electricity, not in cooling the servers. Is most of the water consumption to do with power generation rather than cooling the servers?

If you are going to increase water use efficiency, we have figures about how much water different data centres use per year, ranging from 100 million litres to fewer than 1 million litres. To me, that was a meaningless statistic, because one of them might be answering a million questions a year and the other answering five a year. Do you measure water use efficiency in terms of how much work the data centre is doing? Luisa, since you talked about power versus cooling, maybe you could tell us.

Luisa Cardani: I might not be able to correct some of these estimates. It is worth noting that some of these estimates are looking at the sector globally. Other geographies have different regulations and design choices, which, as the trade association for the sector in the UK, I cannot comment on.

In the UK, where there has been some retrofitting and there have been changes in how you do cooling, the majority of newer data centres are not water-intensive. Therefore, when you do a search, how much water it uses is very difficult to estimate. It will probably be much lower than in other geographies. It is also worth noting that you cannot estimate it, because, depending on the business model, a co-location facility might have 100 tenants; another might have only one. It is hard to estimate at facility level and at search level.

However, it is worth noting that, generally, improvements in efficiency in the sector would significantly reduce any estimate. Steven can probably help with some of those metrics, because the metric that you mentioned is from the Climate Neutral Data Centre Pact.

Steven Campbell-Ferguson: The metric mentioned is the WUE, which has a parallel for power called the PUE, which is a measure of the data centre’s ability to process the heat. I represent a co-location company. I do not know what my customers do with their computers. That is not my business; that is their business. Whether they are doing a million searches, I cannot give you that data.

The WUE and the PUE are measures of what I do, which is to provide the electricity in and get the heat out. That evaporative system is part of getting the heat out. If I am using water, the WUE is how I can measure the efficiency of the system. Those WUEs will depend on the type of system chosen. For that 0.4 litres per kilowatt, I will need to use a hybrid system that operates only when it has to, uses water in peak summer hours only, and operates dry for the whole winter and mid-season. To achieve that, if you are using open cooling towers, that will end up with a WUE nearer to 2 or 2.5, depending on climate and the exact operating temperatures.

That is, in a way, the answer to the earlier question. We assess the availability and the stress, and that dictates some of the technical decisions. Again, we can go fully dry or we can go to that halfway house of getting at least some of the benefit during the peak summer hours, which is where we get the most benefit.

Lord Krebs: Can I just ask for clarification with a yes/no answer? Is it the case that the majority of the water used by a data centre is in generating the electricity to run the data centre as opposed to cooling the servers?

Steven Campbell-Ferguson: The majority used by the data centre is in the cooling system. The power generation is done off-site by power-generating companies. That is not in our control. It is a third party that, effectively, the electricity is coming from.

Lord Krebs: When I read that some data centres use 100 million litres per year, that is cooling the servers.

Steven Campbell-Ferguson: We are evaporating water off the roof to help cool the building. It is the sweating mechanism, effectively, of the building.

Lord Krebs: If I can move quickly to a supplementary, because I am aware of the time constraints, you rely to some degree on mains water supply and to some degree on abstracted water. Should the price that you pay for abstracted water in particular reflect its scarcity? Would that drive greater water-use efficiency? Is there enough of a price signal?

Steven Campbell-Ferguson: At the moment, as John mentioned, the difference between electricity and water pricing makes the water solutions look very economic. If that price changes, we will make different decisions.

Lord Krebs: Is that for both mains water and abstracted water?

Steven Campbell-Ferguson: My experience is that the bulk of water is from the mains. Very occasionally, there is a borehole. I have been involved in one site that did sink a borehole, but it provided only a fraction of the requirement of the site.

John Abraham: I can add some colour to that. It is the town’s water that is the cost. There is a lot of attention being given to the price of water at the moment, with the Cunliffe review and lots of the other reviews of the municipal water industry. Cunliffe has suggested that there should be an escalation for the industrial usage of some water.

This is not a data centre comment specifically, but we were talking to the chemical industry. It would cost them £100 million a year if they were to lose the discount that they get for large bulk water supplies. Whether those large bulk water supply discounts should apply to data centres may be part of it.

If I may just comment on steam versus cooling water use, one of the other biggest constraints about developing data centres is the availability of power. Many data centres are having to go to on-site solutions to produce their own power on-site. They will use steam turbines; they will use water, as they create steam to create energy. The grid has now said that we just do not have enough energy available in our national grid and that we have to start producing on-site. We are working on a few of these plants at the moment and they will use water.

Q114 The Earl of Leicester: I have a very simple question before I get on to my main one. If there is a serious drought and you run out of cooling facility, what happens to the data centres? If they switch off, what happens for the man in the street? Does his internet die? Can he not search for anything on Google?

Luisa Cardani: I can start and I am sure that Steven will have more to say. If there is enough warning and if contracts with their tenants allow, they can migrate workloads. If that is not the case, yes, it is a serious problem. Depending on which data centre and what kind of compute happens in it, it could have a significant impact on the economy. It is worth noting that data centres have been recognised as critical national infrastructure for a reason. As techUK, we are in conversations with government to understand what that means and what support can be put in place, in the same way that you would for other critical national infrastructure such as hospitals and power stations.

The Earl of Leicester: Have we had examples of crashes of data centres?

Luisa Cardani: It is fairly rare that it is for drought reasons. What can happen is that it is not necessarily the drought, but it might be the heatwave that accompanies the drought.

Steven Campbell-Ferguson: The industry has got more mature and learned from experience. Probably 15 or 20 years ago, when we had a heatwave, there was a degree of, “Something is going to happen”. These days, the industry is far more resilient.

The Earl of Leicester: I had better declare my register of interests. I am a farmer in north Norfolk. It is mixed farming, but we also grow vegetables. Therefore, we abstract water from boreholes. The majority of it is from water abstraction from rivers before it flows out to the sea. There are also a number of private households that have boreholes.

What are the main barriers to greater uptake of water reuse? What are the non-potable technologies for data centres’ industrial processes and energy?

Luisa Cardani: I am very happy to start and I am sure that fellow panellists will have more to say. First, treated effluent remains classified as waste. It might involve additional environmental permits for changes in discharge volume, quality and location, and new abstractions. There is a regulatory change that would be needed. There is no formal framework distinguishing between potable and non-potable water demand. Already at the planning stage, data centres often do not really have a choice. The water supplier provides whatever it provides.

Finally, to my knowledge, there are also investment challenges in the sense that some water companies have limits on how much private investment they can receive. Where there could be a joint venture, this is restricted. These are some of the current barriers that the sector is facing at a national level.

John Abraham: If I may broaden it out, away from just data centres, industry uses way more water—copious amounts. For every car that we build, we use 175 cubic metres of water. There is a lot of water getting used. The technology is available for closed-loop solutions where they create not only their own water supply but their own wastewater treatment facility as well. There are many examples where they can then create their own bioresource anaerobic digestion process and their own biogases. They can create circular economy solutions from the waste products that come from those. There are multiple examples of that around the UK.

There is a huge opportunity to do more of that, but the barrier that exists is cost; it is the economics. There is no government funding or grant available for any industry. We held a webinar for industry about the circular economy solutions for water, and how we can alleviate the drought issues and, to be frank, the issues with the municipal water industry, which are extensive, as you all know. The unanimous view was that we need funding; we need some capex.

Back in the day, if you could get some renewable energy, you got ROCs and green incentives, but there is no facility for any industry to get any government help to create a more economical way towards a closed-loop solution in order to make them more competitive in the marketplace and to create and enable growth across the whole of the UK.

One of the points that I wanted to make today is that that funding needs to be made available for industry to grow, because we strongly believe that water and the economics of water is a big enabler for growth.

The Earl of Leicester: What would be the trade-offs of requiring waterless cooling for data centres?

Steven Campbell-Ferguson: The trade-off is that, if we do not use the water, our energy requirement goes up, so we use more power.

The Earl of Leicester: If the price of water increased, it is more likely that, even without government grants, these data centres would invest more in closed-loop systems.

Steven Campbell-Ferguson: Yes, in non-evaporative systems. Closed loop is slightly different, in that we can reject the heat off the roof without using water. Just pure economics would drive decision‑making. There are a lot of issues with water. The risk of drought goes away. We are a very risk-averse business. We want to be operating 24 hours a day, seven days a week, 365 days a year, because we do not want our customers to go down and their customers—the man on the street—to suffer, so there is more energy use.

One other thing that would be a consequence is my peak summer load. If I have a fixed grid connection of, say, 100 megawatts, I can serve around 65 to 70 megawatts of IT load, which is the compute, but I need to keep the other 30 megawatts reserved for my chillers. With water, I do not need to make such a big reservation. As Luisa said earlier, grid connections are a constrained issue in the UK at the moment. I can get more useful power out of my fixed grid supply. There is that benefit as well, in that I do not have to go and build the next data centre as quickly. That is another consequence.

Q115 Earl Russell: Good morning. I will just declare my interest as a non-exec director and a member of the board of the Water Retail Company. It has been a really fascinating conversation. You have said that this is an economic, not a technical, problem. You have made arguments that there should be more grants available to do these things. Can I just ask you about the lifespan of these systems? The chips and all that can be updated, but, once you build a data centre and get it connected, you are there for a number of years, presumably until the 2050s at the moment. Where does this balance lie between grants, cost and leverage? Is it not in the Government’s interest to change that calculation to price you off a dependence on water in the longer term?

Steven Campbell-Ferguson: It depends on what the consequence is. Is it better to invest in the water industry? The UK gets a better return on investing in water resilience than in investing in the grid, so it is a balance.

Earl Russell: There might be a threat of drought in future from climate change. I am wondering whether that could be a technical solution.

Steven Campbell-Ferguson: As John said, yes, there is, but the right investments can mitigate these risks. We can build desalination plants next to the sea. The UK has a lot of coastline and a lot of access to the sea.

Earl Russell: That includes some salinisation and returning saline water back to the oceans, which has impacts.

Steven Campbell-Ferguson: In some respects, we are only doing what the ocean does anyway, which is evaporating water. We just short-circuit that. In the end, it is the same process and the marginal amount of evaporation is tiny.

Q116 The Chair: John, which industries consume the highest levels of water?

John Abraham: You are looking at the food and beverage industry.

The Chair: Is it in that order?

John Abraham: Yes, it is one of the largest. The food and beverage industry is very water-intensive. There are industries that use a lot of water, but we do not have a lot of them. Paper and pulp manufacturing, for example, uses copious amounts of water, but unfortunately—or fortunately, whichever way you want to look at it—we have closed most of our pulp and paper mills over the last 30 or 40 years. Steel is a huge consumer, but, again, that has declined.

In my notes, we are predominantly talking about the chemicals industry and the food and beverage industry. Agriculture is very low, because it has a lot of self-supply and abstraction licences, so it is not taking drinking water in the same way as a lot of other industries. I would say that the food and beverage side and the chemicals industries are two of the largest.

It is also going to change as we evolve into decarbonisation, hydrogen production and carbon capture. They are massively water-intensive. The Government are making available a lot of funding for HyNet in the north-west, for example, for the hydrogen hubs. They are going to use a huge amount of water and will probably overtake everybody else in the amount of water that they are going to need going forward.

Q117 Baroness Whitaker: I have a very quick, simplistic question. I just want to know why the industry needs government help if it is in its commercial interest to use water. Is it not a question of investment and long-term planning?

John Abraham: It depends on the industry. With the greatest respect to the data centre industry, it is very different from some small chemicals industry, or the food and beverage manufacturing industry, using the Earl of Leicester’s products from the fields and doing all sorts of processing. The data centre industry has a lot of money. There is a lot of investment going into it.

Some of the very fine industries that exist throughout the UK just do not have the capex. They just do not have the funding available at the scale that the data centre industry has, so they are struggling to grow because they cannot get any more town’s water. They cannot get any more trade effluent discharge consents agreed. They need to change their processes to take impurities out of the wastewater that they are creating and they just do not have the capex. They are operating on pretty thin margins in the first place. In order to create the right sustainable solution for their site, they just cannot afford the capex.

Baroness Whitaker: We will have to leave it there, but we might pursue this later.

Q118 Lord Lennie: Good morning. Do the data centres and industrial water users share information on water consumption with the Environment Agency or water companies and, if not, why not? I know that you are not required to, but do you share the information?

Luisa Cardani: I am very happy to take this one. They are not required to share the data with the Environment Agency. The reason is that, first, there is no requirement and, secondly, water was not seen as a particular problem for the sector historically, especially in a country such as the UK that has, historically, had a very wet and cooler climate.

Data centres do not have to share their water use data with the water companies, because the water companies bill them, so they do have the data at hand. The question is whether they are able to share it. For example, MOSL cannot share data beyond what the data was collected for, because of data protection, and the same would go for other water companies. The data is out there and is available. It is just not mandated to be shared.

TechUK and the Environment Agency did the very first survey of the sector for England, just to get a bit of a sense and to try to demystify the sector, given the claims that the sector is water-intensive in the UK. Some of the findings echo findings of other European countries that this is not a water-intensive sector in England.

The survey, being the first one, has its limitations, but the idea is to continue and do another one next year, having learned lessons from this one. It is also to move the needle towards better water regulation, so that there is more data sharing between water companies, the regulators and government.

John Abraham: I have mentioned Sir Jon Cunliffe’s review and recommendations. One major recommendation that is pretty much accepted is that they are going to abolish Ofwat, look at some of the powers that the Environment Agency is responsible for, and merge it into a new regulatory group. That will very much facilitate better sharing of data. We are waiting for a response from the Government on that recommendation, but it will help enormously. The EA has a lot of data, but I am not sure that it is all used well enough.

Lord Lennie: There is no lack of data. It is the use of the data.

John Abraham: Yes.

Luisa Cardani: If I may, it is also worth noting that, if you use a dry system, you may not measure your water use, because you have none. Some companies would say that, if there is any water use, it is so limited that they would not feel the need to report it, unless it is internally for their annual report to their shareholders.

Q119 Lord Trees: I am very interested in the geography of this, and I want to be clear. Is it more important for the data centres to be closer to the users, or closer to the essential resources that they require, which you said are energy and water? Both of those tend to be more available in the north and west of our country, whereas the users are concentrated in the south and the east. Where does that balance lie in terms of sustainability, commercial reality and technical ability?

Steven Campbell-Ferguson: It depends on what exactly the data centre is doing. Historically, and if you look at the data, there is a big concentration in the south-east. That stems historically from the finance industry, which drove data centre development initially, and from the City of London. That is why it is the south-east and that is, therefore, where the networks and the data centres are. Data centres want to be near the users, but they also want to be near each other because they share a lot of data. That is historic.

If we look to the new AI workloads and where you are doing AI training, that is less geographically dependent, because, basically, super brains are just thinking on their own somewhere, in a way. When you are coming to the inference, that needs to be back near to the users again to get a swift response, so it is a balance.

Yes, we could shift to where the power is. In the end, based on economics, if the power is available only in the north, we will go there. That means we then have to take the networks to the power as well.

Lord Trees: Is it not easier to shift data than to shift water and energy resources, even? Our grid is straining at the leash at the minute.

Steven Campbell-Ferguson: Again, it is around the time delays. Certain financial industry requirements are very critical on proximity. Someone searching for a cat video has a slightly different requirement. As we see the demand for co-location, our customers generally still want to be in the south-east of England.

Lord Trees: Data moves at the speed of light through fibre, does it not? That is pretty fast.

Steven Campbell-Ferguson: But the trading engines are working on milliseconds.

Q120 Lord Jay of Ewelme: You say that they want to be in the south of England. I can understand why they would want to be, but do they need to be?

Steven Campbell-Ferguson: Do they need to be? They should probably answer. They can manage it otherwise, but they will not be able to provide the service in the same way that their customers may be expecting.

Lord Jay of Ewelme: Could they organise themselves so that they could manage the service in a different way?

The Chair: I guess that there are international considerations.

Steven Campbell-Ferguson: Yes.

Lord Jay of Ewelme: The answer is yes.

Lord Lennie: They could come to Newcastle.

Luisa Cardani: They could. It really depends on what compute is happening, as Steven said. If you have a live feed in a hospital, you want it close enough to the hospital. I am sure that you do not want a live cam in your cavity to have any delay in the ping latency. It really does depend. At the end of the day, most of the sector would say that the customer is king. If you are a customer with a streaming service or a financial service that is able to withstand a certain ping latency, that is where you triangulate your location.

Q121 The Chair: This is of great interest and very relevant. I wonder whether each of you, from your own perspectives, would write to us with further information on the siting and the required level of proximity of data centres to the users. That would be very helpful.

We are just over time, but there is a question that I want to put to each of you, starting with John Abraham. Waterwise raised some concerns in evidence to us about the level of awareness of drought risk within the commercial sector. It suggested that there should be requirements, particularly for businesses using large volumes of water, to have their own drought management plans. What is your comment on that?

John Abraham: If they are taking town’s water directly from the water networks, it is more for the water companies to do that. If they have their own abstraction licences and discharge consents, yes, they probably should, because it would be also a resilience plan for their business continuity, so they would have to. Where you have your own boreholes or abstraction from a river, you would have to have your own resilience planning anyway.

Luisa Cardani: The majority of techUK members, along with techUK, are members of the Climate Neutral Data Centre Pact, so they are encouraged to measure water use effectiveness. However, most of them tend not just to measure water use, but to look at drought and water scarcity in the regions where they plan to build newer developments.

The Chair: You have the last word there, Luisa Cardani. With that, can I thank all three of our witnesses this morning for the excellent testimony that you have provided? We will now have a swift changeover of personnel for the second panel.