Meet our speakers: David Jones
Charging your phone in Lagos produces around ten times the carbon of the same charge in London. David Jones (Pembroke 2025) knows why... He spent 15 years running the infrastructure before coming to Cambridge to map it.
David’s talk on ‘Africa’s Climate Bomb’ will presents his original research, which maps the scale of Africa’s diesel dependency to fuel the continent’s digital revolution.
"Stand next to a tower running on a diesel generator and the fumes work into your clothes and stay for days. Once you have smelled the cloud, you can't unsmell it." David Jones tells us more about his research...
How would you describe your work?
The internet feels weightless, but it runs on a vast physical body: the phone in your hand, the masts on the hill, the fibre under the street, the data centres humming on industrial estates. All of it burns electricity, and electricity has a carbon cost. My job is to map that hidden machine, and put a number on what it actually uses, to give the planners, and the people that are setting global policies, the information they've been missing. This really matters because nobody agrees how big the energy problem is. There are lots of uncertainties around the speed at which AI is growing and how this will impact the demands for energy, plus many of the previous measuring models don’t account for the rising demand for data worldwide. Credible forecasts of the internet’s share of global energy range from about 5 to 40 per cent, an extraordinary spread for something we’re pouring billions into. This huge gap in estimate is driven by uncertainties around internet uptake globally and the differing energy efficiency of the networks.
What initially attracted you to this research?
I've always worked in telecommunications, but for most of that time the energy side was simply a line in a spreadsheet. Then I went to Nigeria and experienced the reality of the situation. Africa is where the future of connectivity is being built, faster than anywhere on earth. You can watch it happen and you can smell it, because Africa’s grid is largely run off local diesel. Stand next to a tower running on a diesel generator and the fumes work into your clothes and stay there for days. Once you have smelled the cloud, you can't unsmell it. There's currently around 1.4 billion people living across the continent with only half of them having regular access to the internet, but in the next 5 to 10 years this is anticipated to increase dramatically (40-50% increase in data traffic annually). Nigerian telecommunications infrastructure consumes around 1.1 billion litres of diesel per year. There are real efforts to make a switch to green energy, but these are overwhelmed by this massive year on year increases in demand, and from vested local interests who financially gain from keeping the diesel flowing. If demand keeps growing without big changes to the energy infrastructure, the diesel cost could be in the hundreds of billions of litres.
What caused the ‘climate bomb’ to become your focus?
I have always been environmentally minded; long before Cambridge I was involved with People and Planet and worked with the United Nations Development Programme. What I came to realise is that telecoms are overlooked by design. The whole industry is engineered to be invisible. We hide the masts, bury the cables, tuck the data centres onto industrial estates, and call the result 'the cloud'. We use words like ‘ether’, ‘virtual’, ‘wireless’, all of which are intangible. Invisibility is the selling point, because the less you think about it, the less pressure there is on telecoms companies to build green. But from the inside, the truth is harder to hide. Our own data showed that barely half the diesel we paid for ever reached the generator. The rest vanished into spillage, theft and guesswork. Combine that waste with explosive growth and a fuel that emits ten times the carbon of a clean grid, and you have a climate problem hiding in plain sight.
David's ADIEM model showing connected devices per km2 in Africa
Your research focuses on geospatial data to map energy use. What can your model show that previous models have missed?
Most forecasts of the internet’s energy usage are top-down, which could be a rough estimation of government forecasts added together, or a prediction based on what data centre companies are currently selling. They take one big number and divide it using broad averages, which hides the two things that matter most: place, and the parts of the world we tend to ignore.
My work is from the bottom-up and spatial. I can now show every connected device on the planet, where the traffic flows, and how much energy that takes, tile by tile. That picture is nothing like the averages. Pull up my map of Nigeria and most of it is dark. The light pools in Lagos, Abuja and a handful of cities. Everywhere else, almost nothing. A top-down model smears usage evenly across that emptiness and never sees the dark at all. It also misses the obvious: a mast out in that interior is almost certainly running on diesel.
The averages tell you the internet is working fine. My map tells you where it is not, and where the cheapest fix sits. It can also show renewable energy investors the areas where there’ll be significant demand for power over the next 25 years and where the biggest impact from a new clean grid will be.
Why is it so important now to make the internet’s physical and environmental footprint more visible?
The climate implications are huge, and the window to shape them is closing. Nigeria burns somewhere around 1.1 billion litres of diesel a year simply to keep its internet running. Now scale that across a continent coming online faster than anywhere on earth and within a decade you are looking at diesel use for digital infrastructure heading towards 100 billion litres. Almost none of this shows up properly in the official figures.
If planners, investors and governments can't see where the energy and carbon sit, they can't steer the build. Once the masts are up and the concrete is poured, the choice is locked in for thirty years. You can't manage what you can't measure, and right now we're building blind.
Newer forecasts on AI energy usage are also quite severe. In the UK alone, within 4 years it’s predicted to go up five-fold, and I think that a lot of these numbers are undercooked. If you look at the amount of money that the hyperscalers like Google, Meta, OpenAI are spending on data centres globally, it adds up to a significant amount. My best guess is that AI data centres will be using somewhere between 3% and 5% global energy by 2030.
“You can't manage what you can't measure, and right now we're building blind.”
How optimistic are you that this trajectory to climate crisis can be changed?
I'm genuinely optimistic, and Africa is the reason. Much of the infrastructure there hasn't been built yet and it's far easier to build something the right way than to rip out and replace something already running. Putting solar on a new tower instead of a diesel generator is cheaper over its lifetime, needs no new technology and needs no grid reform. The opportunity is to lock in the clean choice before the dirty one is set in concrete.
The hard part is incentives. The person choosing the cheapest option this quarter is rarely the person who pays the carbon bill later. One of the things I hope to show with my modelling is that renewable energy projects in Africa can qualify for carbon credits when they’re funding digital infrastructure. Change becomes real when pressure and incentives line up, when doing the right thing is also the obvious commercial thing.
"Industry taught me what the infrastructure really does, the operational truth you only get from running it. Cambridge taught me to be ruthless about evidence and gave me the tools to act on it."
How has Cambridge influenced how you think about solutions compared to your time in industry?
Industry taught me what the infrastructure really does, the operational truth you only get from running it. Cambridge taught me to be ruthless about evidence and gave me the tools to act on it. In industry, your job is to solve the problem in front of you, on time and on budget. You are not paid to ask whether the whole way the sector measures itself is wrong. Cambridge gave me the room to ask exactly that. I am surrounded by some of the best minds in the world, across engineering, energy and economics, and that changes which questions you think are worth asking.
Most exciting of all, it has let me move from only measuring the problem to building solutions to it. I am developing technology that could cut a data centre’s energy use by around a fifth, mainly by rethinking how these buildings handle heat. Computers throw off enormous amounts of heat and stopping them from cooking themselves is one of the largest energy costs in the entire system. If we can accurately predict how data centres will operate under various scenarios, we can better plan when the air conditioning needs to be on, which should save 20% of the data centre energy costs.
If there’s one ‘takeaway’ you want to leave people with from your talk at the Alumni Festival, what would it be?
That it cuts both ways. Left alone, the internet becomes a climate disaster, vast, fast, and hidden behind one soft little word: cloud. Steered well, the same growth is one of the best chances we have to do real good. All that new infrastructure needs power, and that demand can drag clean energy into places that have never had reliable electricity. The internet is not something we have to apologise for, built on purpose, in the right places, it can lift whole regions while cutting emissions. The choice is ours. We just need to see clearly enough to make it.