COMMENT
pay attention to the sustainability question around data centres.
There’s no getting away from the fact that data centres absorb a growing proportion of electricity production, just as there is a broader push towards electrification. If consumer electricity prices go up because of data centres’ power use, that will feed through to unease and resentment. A Manhattan-sized data centre on anyone’s doorstep is hardly going to go unnoticed, all politics is local after all.
This should mean the sweeping statements we’ve seen from political leaders over the last year or so start turning into real world policies. These need to fuel real world progress in turning around planning holdups or speeding up electrical hookups just so builders can break ground. But we should also expect more detail on what data and AI sovereignty really means, and how this affects what is built where.
Broader geopolitical questions will become more pressing in 2026. The White House is keen to keep the US’s AI edge, and has already restricted NVIDIA exports to China. More recently, President Trump has mused that Jensen Huang will have to keep his top end chips exclusively for US users. It’s impossible to predict what will happen for sure, except to say that whoever is in the Whitehouse has more bearing on what you have in your data centre than at anytime since the early 1990s. [Yes, datacentres did exist in the early 1990s. And so did export controls.]
All of this suggests ever more uncertainty around the data centre industry. But I prefer to look at the upside, which is that these pressures and uncertainties will also drive innovation.
We know that ever faster GPUs will suck in ever more power and throw out ever more heat. That will demand innovation in cooling, specifically liquid cooling. The mega data centres on the drawing board at Meta will demand cooling solutions that are highly efficient, but can also be built out quickly and easily, as will the new facilities being built in the Middle East. Those edge sites will demand flexible cooling architectures that can be adapted to specific locations – yet do not leave operators constantly reinventing the wheel.
So, not only will we see increased take up of liquid cooling, but we will see a shift to modular, scalable systems that can be easily adapted to a myriad of locations. Think standardised technology that can be easily tuned for multiple installations, rather than bespoke and expensive.
And the heat they remove will still have to “
Last December, it might have felt like the last few years of breakneck growth were transitioning into something slightly more sedate. Instead, things just seem to have accelerated. More mega investment plans announced, faster GPUs unveiled, more government interest in the sector.
go somewhere. So, brilliant engineers will focus on novel, useful ways to reuse the vast amounts of heat our increasingly vast datacentres are going to be producing. The case has already been proved with district heating projects, whether at Queen Mary University of London, or Stockholm Data Parks. And there are plenty of proof cases for reusing excess heat for hot water and heating swimming pools.
”
But we should also expect more initiatives around reusing heat for agriculture, and increasingly for a wider variety of industrial processes. Lighter industries at first, but hyperscalers’ sites will be producing heat on a truly industrial scale, so we need industrial processes to match.
With great power comes great responsibility
We all know that data centres are inextricably intertwined with society at large. But as the importance of data centres becomes more apparent, something else happens. Businesses, governments and citizens begin to understand that data centres are, indeed, essential critical infrastructure. That means vindication for those of us who’ve been arguing exactly that for years. But it also means responsibilities. So, we should prepare for growing awareness of the need for resiliency, standards and reliability both at the broad data centre level and for the specific technologies within them such as liquid cooling.
As data centres step further into the public eye, our role as the backbone of the digital economy is no longer in question. With that visibility comes scrutiny - from governments, regulators and the public alike. Rather than resist it, our industry must be ready to explain its value, and show how we’re enabling progress, sustainably and transparently.
So, my biggest prediction is that we will see a more mature, more responsible and more engaged industry. One that doesn’t turn away from its responsibilities but embraces them.
One that brings the same innovative, methodical, engineering-led approach we’ve relied on to get here to solving the inevitable challenges that being “here” creates.
Because the alternative is not sustainable. Not next year, and certainly not in the years to come.
DECEMBER/JANUARY 2026 | ELECTRONICS FOR ENGINEERS 7
Page 1 |
Page 2 |
Page 3 |
Page 4 |
Page 5 |
Page 6 |
Page 7 |
Page 8 |
Page 9 |
Page 10 |
Page 11 |
Page 12 |
Page 13 |
Page 14 |
Page 15 |
Page 16 |
Page 17 |
Page 18 |
Page 19 |
Page 20 |
Page 21 |
Page 22 |
Page 23 |
Page 24 |
Page 25 |
Page 26 |
Page 27 |
Page 28 |
Page 29 |
Page 30 |
Page 31 |
Page 32 |
Page 33 |
Page 34 |
Page 35 |
Page 36 |
Page 37 |
Page 38 |
Page 39 |
Page 40 |
Page 41 |
Page 42
