Should AI Data Centre Roll-Out be Phased and Linked to Renewable Installation?

In the dreamy and often overlooked backwater of Dumfries and Galloway, on the site of the former Chapelcross Nuclear Power Station near Annan, an AI Data Centre is being proposed. The company CX wants to build a 55-hectare, 1 gigawatt (GW) ‘net-zero technology campus.’

There is clear evidence that at least some locals are not happy. The Dumfries and Galloway Standard of 2nd October 2026 reports that ‘Campaigning Dumfries Residents are determined to stop [the] proposed £16.5bn hyperscale data centre.’ They go on to say that ‘A petition against the plan has already attracted more than 1,000 signatures.’

What’s more, discrepancies are alleged between what CX have actually proposed and what CX are suggesting in their literature. So, the Standard reports the ‘Say No’ campaign arguing that the development is ‘on agricultural fields, despite what the developer is trying to say.’

What then is the truth? A Community Information Session is planned for the 13th October which has the intention of providing answers to all such questions. All are welcome.

AI—Regulatory Concerns

Data centres are fundamental to the further development of Artificial Intelligence (AI). They have proven to be unpopular in the United States, as is made clear in a recent report from the United Nations University in Tokyo. In summary this unpopularity is because ‘The centers cover large areas of land, require extensive amounts of electricity to operate and need substantial water to cool the equipment, raising concerns about their impact on the environment and local electric bills.‘ [1]

However, the entire AI project has come under scrutiny in recent times with regard to the possibility that the technology may go rogue with developers losing control unleashing an entity which may harm rather than help humanity.

In a recent BBC interview with Faisal Islam, Jack Clark has suggested that there is a need to ‘put the brakes on Artificial Intelligence (AI).'[2]

Clark is the co-founder—together with six other individuals—of the AI company Anthropic.[3] His central concern is not the energy demands of data centres. Rather he suggests that there is a windfall available to us all from the current iteration of AI development. This can be harvested now to our considerable advantage. He also argues, however, that there are significant concerns about further AI development that require reflection, a matter which has indeed been discussed elsewhere on this blog. [4]

Anthropic has shown a willingness to consider the risks and the difficult moral questions which are posed by a technology that threatens to out-think us all and yet potentially be available to the most unscrupulous in our midst. On February 24th this year, Pete Hegseth, the US Secretary of War —as he styles himself— gave an ultimatum to Anthropic: if it did not agree to terms set by the Pentagon on usage of its AI for military purposes, it would face severe penalties.[5] It was precisely Anthropic’s questioning and thoughtful approach which triggered Hegseth’s impatient response.

There are other versions of AI on offer. They are all in frantic competition for the power and prizes that AI offers to whoever gets there first. The question inevitably arises as to whether this rush for glory can be paused. It is in the context of this gold-rush hysteria that the proposal of a ’55-hectare net-zero technology campus’ is raising concerns in Dumfries and Galloway.

CX Information Events

CX have been staging events to inform the public of exactly what is planned. I attended one of these in June, with some trepidation. I had been expecting a full house of disgruntled locals and a beleaguered panel of experts attempting to answer questions based on a fairly broad spectrum of information and misinformation. As it turned out, when I found myself in Victoria Halls, Annan, shortly after 4pm on Wednesday 17th June 2026, I wondered briefly whether I was in the right place.

The room was spacious. There were some panels of information. There were perhaps a dozen people milling about with a few coming and going during my visit. Annan is just a short distance from the development and I had expected more interest. Perhaps the turnout had been greater at the previous day’s event at Creca Village Hall. Creca is the nearest small community to the proposed new campus. Indeed, I had seen a Facebook page which suggested some fairly consolidated negativity to the whole enterprise from residents in the immediate vicinity of the proposed campus. But on the whole, what I observed in Victoria Halls was mild-mannered curiosity.

I was greeted at the door by a lady who assured me that I was in the right place. She asked me to sign in and wondered if I was from the press. Perhaps it was the pen in my breast pocket which suggested this, or maybe just my shady appearance or my shifty manner.

‘No, no!’ I reassured her. ‘I sometimes write a blog.’ ‘Oh blogs are all right!’ she smiled. I wondered whether being a journalist might have upgraded my status—or, conversely, got me shown the door.

Experts

The lady then pointed out a few people whom she described as ‘experts.’ I wandered off in hopes of speaking with one of them. I looked for badges with names and job titles. There were none. I reported back to the lady on the door and she told me that the three experts were David, Kyle and Carly. Carly had just gone out for a moment. I wrote the names down and asked for surnames. I was politely told they don’t provide them. Given the realities of modern social media, I understand their reluctance.

As I was chatting to the lady at reception, Carly returned and I was introduced. I spent the best part of three-quarters of an hour talking to Carly. I found her a sympathetic and attentive listener who seemed to understand the concerns I raised. She did her best to address those concerns, and when she did not have an answer, said so and suggested that I might ask one of her colleagues.

Planning Process

Carly told me that she was a planner, based in London. I took the opportunity to quiz her on the planning process in this case.

What CX Technology is currently seeking is Permission in Principle. This amounts to an outline proposal and will not be sufficient to grant permission for the project to go ahead. If this preliminary application is successful, CX will then be able to submit a detailed planning application on the basis of which the local planning authority may green-light or perhaps refuse the application.

Carly seemed to think the detailed plan might be ready by next year. There would be ample opportunity at that stage for the public to look at the detailed plan and to question it.

Where will the proposed data centre get its electricity from?

Would renewable generation available in Dumfries and Galloway meet the requirement of the data centre, and still leave anything left over for other purposes?

Carly said that the scale of the data centre was not necessarily 1 GW. I asked what the minimum scale of the development was. Carly referred me to David for further detail on this matter. When I spoke to David about this, he told me that the construction of the Data Centre would be phased and each phase would be 240 MW. The CX plan was a four-phase development which would effectively be completed as a 1 GW plant (or more precisely, 920 MW).

David suggested that developing technology may in practice reduce the energy demand of the data centre without compromising its processing capacity. I argued, however, that a 1 GW plant would in all likelihood consume most of the renewable energy generated in Dumfries and Galloway. David said that the availability of renewable energy was a key reason for building in this area. I pointed out that the Data Centre would compete with energy customers in Dumfries and Galloway for example, charging electric vehicles, or running heat pumps to warm their homes. He suggested that there would be no conflict. This is a question I look into more deeply in the section below, Some Rough Calculations.

Impact on the Price of Electricity

I started to focus more on the impact of the Data Centre on the price of energy locally. David pointed out that the price of energy in the UK, currently, is not set locally. I was aware of this point, but I am equally aware that companies like Octopus Energy offer very attractive— sometimes free—deals to those who charge their electric vehicles at off-peak times. I can’t see how there would be off-peak times any more if we have a data centre.

I observed that Greg Jackson of Octopus Energy has argued for local pricing. [6] David’s response was interesting. ‘That wouldn’t be very fair on people in London, where there aren’t wind farms.‘ I pointed out that wind farms aren’t universally popular in Dumfries and Galloway, and perhaps there should be more wind farms in the south of England. ‘Not much wind,‘ I think he said.

In our discussion, David mentioned the figure of 3 GW as being the generating capacity of Dumfries and Galloway renewables. I expressed skepticism regarding this figure. After all, renewables cannot consistently generate at full capacity.

A well-informed friend had suggested 1,700 GWh as a reasonable current estimate of actual local wind farm generation. However, David’s figure of 3 GW generating capacity is quite close to the 3.2 GW my friend suggested is projected for 2035. Just to be clear: these latter figures refer to generating capacity. Actual generating output depends on variables such as wind and sunshine.

Lots of ifs and buts there, not least the average output—circa 1 GW—arising from the potential capacity ~ 3 GW.

Bespoke Renewable Generation

Carly was quite clear that the development would not include any bespoke integral wind or solar development, but would rely entirely on existing and projected local generation.

I also discussed energy costs with Carly, mentioning that data centres have proven very unpopular in the United States. I suggested that the local cost of renewable energy would be likely to rise simply because of the competition for energy introduced by the Data Centre. The AI companies behind the data centre will have no problem paying the higher prices and are unlikely to be concerned about the possible impact on energy costs locally. Carly was, I would say, more circumspect in her response to this proposition than David had been.

Some Rough Calculations

According to UK Department for Energy Security and Net Zero (DESNZ) onshore wind in Dumfries and Galloway currently generates ~2,127 GWh (gigawatt hours) per year. [7] Obviously daily generation of wind turbines varies significantly but using the DESNZ figures, that’s an average output of 5.82 GWh per day. The expectation is that, unlike wind generation, the data centre will, typically, continuously operate close to its full capacity. A 1 GW data centre operating at full capacity for 24 hours would consume 24 GWh per day—nearly four times the average daily wind generation from the entire Dumfries and Galloway region.

There are other renewable energy resources available within or adjacent to the region such as the wind farm at Robin Rigg and the hydro scheme on the River Dee. But even allowing for capacity factors and variable generation, these rough calculations suggest a significant shortfall relative to the demands of the plant. The Data Centre would make up this shortfall from the grid by using fossil-fuel-generated electricity.

We must hope that if CX obtains Permission in Principle, that their detailed application will provide clarity regarding the probable impact of the proposed data centre on the region’s electricity supply and electricity prices.

Gas and Biomass as Energy Sources for the Data Centre

I also asked Carly about biomass. She said there are no plans to generate energy for the campus using biomass. I pointed out that the CX Technology Campus website suggests that biomass is indeed part of the plan. With the help of Kyle—who I learnt was in charge of comms— we checked. There it was. Biomass, clearly suggested as a possible source of energy for the development. But Carly and Kyle were clear. That was an error and the website would be updated to reflect this.

At the time of writing, the update has still to be made.

As regards energy generated with hydrogen: I didn’t really get any clear information about the source of the hydrogen, other than that hydrogen production on site would not form part of the development. So would it be green hydrogen (renewable), blue hydrogen (fossil-abated), or grey hydrogen (fossil-unabated)?

Those questions were not answered other than by the commitment to ‘as close to net zero as possible.’ This commitment might suggest the green option, but that needs confirmation. And then there is the further question of the competing demands for limited supplies of green hydrogen. Green hydrogen is, in general, a product in short supply. [8]

Cooling

How is the data centre to be cooled? Here the answer was straightforward and encouraging. Water cooling for the Data Centre would use a closed loop. Carly said this is standard across the UK and pointed me towards the QTS scheme in Northumberland as an example. No water, Carly assured me, would be routinely drawn from the River Annan. [9]

Jobs and Training

Carly was uncertain as to the number of jobs which might be created by the CX Technology Campus. I mostly expressed interest in the operational workforce as distinct from the construction workforce. On this question the Dumfries and Galloway Standard reports CX’s more detailed claim that the plant will support ‘540 once operational plus hundreds of indirect posts.’

This figure seems remarkably high given that AI threatens to be a grim reaper of employment in the future. Clearly more detail is needed. How many of these proposed jobs are managerial? How many are skilled technical positions? How many relate to security for the plant? How many are unskilled and most probably low-paid jobs related to routine care and maintenance? And finally, what really is meant by ‘indirect jobs’?

Conclusion

These questions, however important, are a distraction from the realities of our rapidly changing world. If you haven’t already noticed, AI is coming at us like an express train. Data centres are part of the package. The CX Technology Campus, at least in the basics of its presentation, feels quite benign. And yet I can’t overlook the fact that, unless I have erred seriously in my rough calculations, it will swallow up most of the renewable energy in Dumfries and Galloway. There remain significant questions regarding the source of energy necessary to provide baseload.

Were I to be entirely in control of the matter, I would say that, obviously, we should not be building this kind of thing until we sort the climate crisis at a more fundamental level.

However, the reality is more likely to be that standing against the construction of data centres in general will be like King Canute trying to hold back the tide. A better way may be to work with that tide; to hold the developers to account; to hope that we can negotiate our way through to better solutions to all of these problems.

One way in which this might be done would be to stage the rollout of the Data Centre. Thus, rollout would be in increments clearly linked to the installation of renewable resources to match the energy consumption of each stage.

CX—according to the Standard—has submitted plans for eight 120 megawatts(MW) data centre buildings with a total capacity of 960 MW—so, close to a gigawatt. Just one of these buildings would be a significant development. Additional buildings could then be permitted when the necessary renewable resources had come on stream.

No doubt, to work effectively, this link between renewable installation and Data Centre roll-out would require a national strategy, but putting such policy in place is exactly what our governments are for. Were the renewables link to be embedded into AI rollout, we might then get virtuous feedback with AI developers prioritising renewable development and installation as a necessary partner of their project.

The challenges of this strategy are, no doubt, very great. Yet technological solutions pertinent to these challenges do seem to be in touching distance.

Perhaps, for example, it is just possible that improvements in AI will accelerate our progress in resolving the problems that are currently preventing the construction of nuclear fusion reactors at scale. That could mean cheap, plentiful, clean energy absent the problem of nuclear waste posed by fission reactors.[10]

Or at a more prosaic level, I have been following the efforts in the UK of Standard Gas. They have been trying to raise finance to rollout their SG100 plants. The SG100 Advanced plant transforms waste into a clean biogas or synthesis gas and biochar.[11] I am thinking there should be an SG100 installed in every waste tip in the country.

There are many other examples of similar technologies in development and therefore many reasons to be optimistic.

Surely, we owe it to the future to be hopeful.