Are Data Centres the Problem, or Just a Phase in the Technology Cycle?

Data centres have become one of the latest symbols of modern energy consumption.

Every search, every streamed film, every cloud-based file, every AI prompt and every digital process has to run somewhere. Behind the clean simplicity of a laptop, phone or app sits a very physical infrastructure of servers, cooling systems, backup power, security, buildings, grid connections and water use.

The growth of artificial intelligence has made this more visible. AI does not feel like an industrial process, but it is. It uses electricity, it creates heat, and that heat has to go somewhere.

In simple terms, nearly all the electricity used by a data centre ultimately becomes heat. The processors, servers, drives, power supplies and cooling systems convert electrical energy into thermal energy. That heat then has to be removed from the building and dissipated into the environment.

This is where the environmental debate becomes more complicated than just counting kilowatt hours.

A data centre is not only a large electricity user. It can also be a large heat source. In some locations, cooling can involve significant water use. In others, the release of heat to the local area, or the indirect pressure placed on waterways and water infrastructure, becomes a planning and environmental concern. Data centres may look clean from the outside, but they are effectively industrial energy conversion sites.

The obvious question is whether that waste heat can be used.

In some cases, it can. There are already examples of data centre heat being used to warm swimming pools, leisure centres, homes, offices and district heating networks. This is a much better approach than simply rejecting the heat to atmosphere or water. If a data centre is going to produce low-grade heat every hour of the day, it makes sense to ask whether a nearby building has a matching heat demand.

But we should also be realistic. Heat reuse is useful, but it is not a complete answer. The heat is often low temperature, it may need upgrading through heat pumps, and the data centre has to be located close enough to a heat user. A swimming pool, hospital, university, apartment block or district heating network may make sense. A remote out-of-town data centre beside a grid connection may not.

The bigger question is what data centres will look like in 10, 20 or 30 years.

That is where I think the current debate may be missing something important.

We are judging the future of data centres based on the physical scale of today’s technology. That is understandable, because today’s data centres are very real, very large and very energy intensive. But technology rarely stands still, especially electronics.

Think about what has happened over the last few decades. Computers that once filled rooms became desktop machines. Desktop machines became laptops. Laptops became tablets and phones. Processing power that once required specialist equipment is now carried in a pocket.

There is no reason to assume that data centres are immune from the same direction of travel.

The demand for data, storage and processing is clearly going up. AI, automation, digital twins, smart energy systems, video, connected devices, business software and cloud services are all pushing in one direction. More data. More processing. More digital dependency.

But the technology doing that processing is also moving. Chip manufacturers, data centre operators, cooling specialists and software companies are all working on the same problem: more computing power, less energy, less heat, less space.

This includes more efficient processors, specialist AI chips, liquid cooling, immersion cooling, better power management, improved server utilisation, edge computing and potentially, in time, quantum computing or other radically different computing architectures. Not all of these will arrive at the same speed, and not all will deliver what is promised, but the direction is clear.

The pressure to reduce energy consumption is not just environmental. It is commercial.

Electricity is a major operating cost. Grid connections are becoming harder to secure. Cooling is expensive. Water use attracts local opposition. Planning is becoming more difficult. If a company can deliver the same computing power from a smaller, cooler, cheaper and more efficient system, it will.

That is why I think today’s very large data centres may eventually be seen as a transition technology.

Not because digital demand will fall. It will not.

But because the physical infrastructure needed to provide that digital service may change dramatically.

Will a data centre that fills a large industrial building today fit into a plant room in 10 years? Could some of that same processing capability fit on a desk in 20 years? It sounds ambitious, but history suggests that electronic systems often shrink faster than we expect.

The competing forces are fascinating.

On one side, data use is increasing rapidly. On the other side, processing efficiency, chip design, cooling technology and computing architecture are also improving rapidly. The future shape of the industry depends on which force moves faster.

My guess is that data centres will not simply keep getting bigger and bigger. They may become smaller, more efficient and more numerous.

Instead of a smaller number of massive energy-hungry sites needing huge grid connections, we may see a more distributed model. Smaller data centres located closer to where the data is used, closer to where heat can be recovered, closer to local renewable generation, and designed at a scale that is easier to connect, cool and integrate.

That could be better for the grid. It could be better for planning. It could be better for heat recovery. It could also make data infrastructure less visible and less controversial.

This does not mean we should ignore the energy demand of data centres today. We should not. Large new electrical loads matter. Water use matters. Waste heat matters. Grid constraints matter. Carbon emissions matter, especially if backup generation or new fossil fuel generation is used to support growth.

But we should be careful about assuming that today’s data centre model is the permanent end state.

The AI energy debate often sounds as though we are building a fixed future: more AI means more data centres, more grid strain, more water use and more heat. That may be true in the short term. But in the longer term, technology development may completely change the shape of the problem.

The better question is not simply, “How much energy will data centres use?”

The better question is, “What will a data centre be?”

Will it be a warehouse full of servers?

Will it be a modular unit beside a leisure centre providing both computing and heat?

Will it be a small local facility serving a town, a hospital, a university or an industrial estate?

Will it be built into the energy system as a flexible electrical load and a useful heat source?

Or will the processing power of today’s data centre be compressed into something far smaller by the next generation of chips and computing technologies?

The public concern is understandable. A large data centre proposed near a town, river or constrained grid connection should be properly questioned. But we should not treat data centres as the new permanent enemy of energy and climate policy.

They are part of a technology cycle.

They are energy intensive now because we are in a period of rapid digital expansion using today’s hardware. The same market that is creating the demand is also creating the pressure to solve the problem.

Use less power.

Create less heat.

Use less water.

Recover more heat.

Fit more processing into less space.

Connect more intelligently to the grid.

That is where the future is likely to go.

So perhaps the message is not “do not worry about data centres”. We should worry enough to design, regulate and locate them properly.

But we should also recognise that the massive data centre of today may not be the data centre of the future.

In 20 years, we may look back at some of today’s data centres in the same way we look back at early computers: impressive for their time, essential to progress, but large, inefficient and quickly overtaken by the next generation of technology.

The future of data will be bigger.

The future of data centres may be smaller.

While You Are Here…

Why not use one or all of Bryce Energy Services free online calculators.

Carbon Footprint — Solar Panels — Electric Chargers — ISO 50001 Gap Analysis

Or demonstrate your business is environmentally and socially responsible, with the CSR Accreditation.

Try our ‘new’ New Vehicle Carbon Footprint Calculator now.