How AI’s electrifying the energy transition with Luminus’ CEO Grégoire Dallemagne

Energy independence is no longer an abstract policy goal when your region imports most of its oil and gas and global tensions spike prices overnight. We sat down with Grégoire Dallemagne, CEO of Luminus, one of Belgium’s largest electricity producers and energy suppliers, to talk about the fastest path to lower emissions and stronger sovereignty—electrification. We walk through the simple physics behind electrification, from why electric vehicles use far less energy per mile than thermal cars to why heat pumps outperform gas boilers, and how shifting more of daily life to electricity can shrink total energy demand. 
 
Then we get practical about AI. Electrification sounds straightforward until you look at what it does to the grid and to customer expectations—smart metres create real-time energy data, renewables add weather-driven variability, and people want prices and guidance that match the moment. We unpack how Luminus uses AI and machine learning for renewable energy forecasting, how smarter models help integrate wind and solar into the electricity system, and how AI-enabled tools support flexibility like smart EV charging that’s cheaper for customers and healthier for the network. 
 
We also tackle the tough questions. AI uses electricity, so when is it worth it and how do you keep “garbage in, garbage out” from undermining results? Grégoire explains why solid processes and data governance matter, how internal AI governance supports safe adoption, and why the standard for customer experience is rising fast with large language models (LLMs). If you care about AI in energy, the Belgium energy market, grid flexibility, or the future of electrification in Europe, this conversation gives you a grounded, operator’s view. Subscribe, share with a friend in energy or tech, and leave a review with your biggest question about electrification and AI. 

Join us and listen to all episodes on www.pwc.be/aiunscripted

Transcript

Welcome And What Luminus Does

Speaker 2 0:03

AI unscripted season two. Today we have another uh video, and today we're gonna talk about Luminus. Luminus, one of the largest electricity producers and also energy suppliers on the Belgian market. And today I have a discussion with um Grégoire Dallemagne, the CEO of Luminus.

Speaker 1 0:24

Welcome, thank you, and um very nice to be with you for this conversation today. Thank you, thank you.

Speaker 2 0:36

Maybe for the people that don't know you or Luminus, maybe indeed a brief summary of who you are.

Speaker 1 0:44

Sure, my name is Grégoire Dallemagne. I'm the CEO of uh Luminus, and at Luminus um we we are an electricity producer but also an energy supplier. We also provide energy efficiency solutions. All together, our 3,000 people are mobilized to build a CO2-free energy future thanks to electricity to help reconcile protection of the planet,

Why Electrification Cuts Energy Use

Speaker 1 1:10

uh human well-being, and economic development.

Speaker 2 1:13

Okay, and the link with AI. Some people might say we don't understand the link with AI. Why did you invite Grégoire? Well, because I know what the link is, but maybe what's the link with AI and uh start off with it.

Speaker 1 1:26

I think these days, um I mean the difficulties we observe in the Middle East, uh, which is which is just uh pure uh horror in a way for many people. Because we are filming now beginning of March. Right, but that reminds us as well that for uh Belgium and Europe who do not have oil and gas in their soil, it's really important to reduce the dependency on the import of oil and gas. And how do you do that? Well, a big part of the answer comes through the electrification. And um why is that? Maybe before we go into the AI uh thing, there's the first reason is when you electrify, you reduce the total consumption of energy. So if you uh drive one kilometer with an electric car more than with a thermal car, you divide the energy consumption by three per kilometer because of the higher efficiency of the electrical engine. Uh if you warm, if you heat a home with a heat pump, you consume you consume three to four times less energy than if you warm the same building with a gas boiler, for instance. So the first reason why you need to electrify is to reduce the total consumption of energy. The second reason is it's the only energy vector that you can produce on your soil. So in Europe, we don't have or almost don't have oil and gas under uh our soil. So for the strategic independence? Absolutely, and so electricity is the only one you can produce in Europe, and this way reduce importation of oil and gas, which of course makes you dependent of um, let's say, one or another uh regime, or of tensions on oil and gas markets. So that's the second reason, and so electrifying is essential to reduce the total energy consumption, to increase sovereignty, and to reduce greenhouse gas emissions, because obviously electricity is also the only energy that you can produce at industrial scale with very low CO2 emissions, which is the case with nuclear, with um wind, uh solar, and even with uh gas-fired um uh production of electricity. Okay. And

The AI Link: Complexity Explodes

Speaker 1 3:55

so, what's then the link with AI? And then the link with AI is that this electrification story is very exciting and it comes with an exponential complexity. Um I can give you a few examples in a second. And AI, um the reason why we all at Luminus embark in AI is to help us precisely deal with this complexity because what we want as Luminous is to make the energy transition something simple and affordable for everyone in Belgium, and AI will help us do exactly that.

Speaker 2 4:32

And before we go to exactly the examples, I think in terms of AI, typically when I talk to CEOs also in the previous sessions, there are two aspects. First of all, it's getting the people within the organizations jumping on an AI train if it's now to become more efficient internally or to indeed be uh more effective. And the second dimension is indeed what you are saying, looking to the market, looking at developing a new business. Right. If we start with the first thing, uh maybe it's not the most important one, but just to to understand a bit how Luminus is dealing with that. How are you transforming Luminous internally itself with with AI? Is that one of the hot topics and how are you doing it? How are you bringing everyone on board? Is everyone using AI? Does everyone have access to AI tooling and so on?

Speaker 1 5:27

I would say when it comes to strategy and organization, we've gone through a number of waves of um uh tuning of the strategy and because and the organization and the processes in the last years because of uh, among others, external circumstances, um, a very long period with low electricity price, energy crisis, COVID. So that has helped us shape the strategy we have today. And as such, AI is not quite changing the organization of uh Luminous today. What it will help us with is

Smart Meters Create Real-Time Data

Speaker 1 6:05

deal with an exponential complexity of what we are doing. Let me give you one example. Uh, a few years ago, in in Belgium and around the world, actually, um, we would do um one meter reading per year. So if you consume electricity in your home, once a year someone will come and look at the counter and say how much did you consume uh electricity in one year? And after that, the the grid operator would send the data to the supplier, and the supplier would give you a bill. One data point per year. What is it today? Uh we have real-time data. So um grid operators have rolled out digital meters, which is a great thing. It's very necessary. We support that uh for the sake of clarity. And so you can know your electricity consumption constantly. It opens a new world of opportunities. First, it comes with an explosion of data. Secondly, you can decide to consume electricity when electricity is abundant and cheap. Um, you can load your car on a Sunday afternoon when it's uh sunny. And these become new standard expectations of customers. Now, for Luminus to deal with this explosion of data, new services, flexibility, and all other examples we can talk about, AI really makes today the difference to models. Do exactly that. Well, in a number of uh a number of uh situations. The first thing is Luminus is one of the leaders in the production of renewable in Belgium. We are leader in wind onshore production, we're one of the large installers of solar panels. We uh also produce hydroelectricity. So renewable

Forecasting Wind And Solar With AI

Speaker 1 7:55

is our core one of our core business. Now, when you integrate more renewable energy uh in the electricity network, it comes also with uh challenges because you need to do the right weather forecast. And so we have an entire team of weather forecasters, and these guys, for example, use AI to forecast production better. So, how much uh solar electricity will our customers produce today, how much wind electricity will we produce, and we go really, really, really granular in the analysis, and we build always better models to have better forecast, and that helps us integrate more renewable into the system. On the other side, customers like to know when they need to do some task and when they need to consume before no one would care. And so today we also uh provide apps that helps you, for example, smart charge your car. And smart charge means you charge at the right moment, so it's good for the network, it's good for yourself because it will also end up being cheaper. Just to give one example of what we do with AI.

Speaker 2 9:06

Okay, but that means that of course you you now have people profiles on board that you may not have had one, two, three years ago because these are capabilities that are now of a different nature.

Speaker 1 9:19

I would say it's just that it's taking importance. It's not that we were not doing all sorts of data related stuff before. Um, as such, large language models are now showing up. Um, the importance of data analysis is even increasing, but it not it did not appear overnight. So I would say the the importance is probably growing, and then these uh people are today formed on new tools, but the strategy is not completely changing. Now I met recently with our team who is doing

Smart Charging And Customer Flexibility

Speaker 1 9:55

a data and AI kind of application in our optimization department, talking about forecasting, and um I was really amazed by the let's say the energy of that team and the ideas they have and the solutions they come up with, and it just goes so fast. So, what we try to do is to provide the right framework for all our colleagues in this is one of the domains, there are many other uh domains, to actually really leverage the use of technology to deliver our strategy more efficiently, with a better outcome and with more impact.

Speaker 2 10:35

And that's interesting, yeah, because if you look at we've done a number of surveys, the Global CEO survey is one of those that we've um launched just after new year. And there's a bit of a mixed pack in terms of feedback, and it's those groups that are looking at AI mainly in terms of efficiency, so looking at the back office, the backbone, and are they able to use AI to create efficiencies? There they seem to be a bit disappointed for the time being on the return on investment. Your story, of course, is a completely different one because you're looking at your core business as such, and and there you are incredibly optimistic, or you sound very optimistic.

Speaker 1 11:14

Yes, and I am, and I am, and um AI, and it's also interesting to bring everyone along because our people feel that AI can help them make the difference in a way AI augments them. Um so they can do more complex tasks, they can deal with larger data sets and come with value added faster with more relevant um uh activities. So it really helps them um while making them and us luminous more successful, but in the end to deliver a better service to the customer and to many stakeholders, among others the grid, and so I would say society at large.

Speaker 2 11:59

But data scientists you already had in the past, you are you then now trying to hire well, we probably need rather more of them uh than less.

Speaker 1 12:12

Uh also we see more and more that IT is becoming um a skill set uh that is more spread across the organization. So we have and we need to have a strong centralized IT team, but the role of the business is probably getting closer to IT and vice

Adoption Inside Luminus And New Roles

Speaker 1 12:35

versa than it was um 10 years ago or even longer when I started my job at Luminous. At the time, those jobs and IT and business departments were more apart from each other. These days we see more of a convergence and we see IT and business departments being as partners in a story where we build together a value added, where we build together new use case to make customer life more easy, a forecast more accurate, to reduce the imbalance on the grid. Collaboration becomes even more important.

Speaker 2 13:14

And did you then set up a specific governance related to AI? What you can do, what you can't do.

Speaker 1 13:20

Yeah, actually we did. Um I mean the thing everyone remembers is the sort of um coming of ChatGPT. Um and uh ChatGPT really changed uh quite many things. And when ChatGPT celebrated its first year, I think on that very day we had a very large management meeting. At Luminus, we bring like 500 uh people together, and the team was AI, and it was um a nice occasion, first birthday of uh ChatGPT. And at that moment, we launched our program

Kairos Program And AI Governance Rules

Speaker 1 13:55

which we called Kairos. And and because in the Kairos you had the A and the I of AI, and Kairos also meant in Greek, not an expert in Greek, but it's sort of the right moment. Okay, and it was we thought it was the right moment to go full speed on um the rollout of this uh new technology, also to encourage adoption across the entire company, and in a way democratize it. So, on the one side, we had our responsibilities, and our CIO at the time and a few business sponsors came with some sort of framing on what is it you do with AI, what you don't do with AI, sort of rules of good use of AI. And so putting that framework to the benefit of the organization, and at the same time encourage everyone, just everyone, to then use it to see how it could make a difference in their process, in their problem they wanted to solve, to make, for example, our customer life easier, nicer, uh in a world where complexity goes up.

Speaker 2 15:05

Yeah, because it is indeed complex, because um it's maybe a bit provocative, but of course the the use of AI itself is of course also a massive energy drainage. True. So that is of course also something that I guess you're also thinking about.

Speaker 1 15:20

Yeah, and I'm concerned by that. Um honestly, we see the use of AI is a I mean comes with a high electricity consumption, yeah, and we might argue is this for good or or not? And it's a fair conversation to have. When you look at our business, the challenge we face today as society with a big S is

AI Power Use Versus Fossil Dependence

Speaker 1 15:45

that the vast majority of energy we consume today is oil and gas. If you look at the energy mix in Belgium, 75% of the energy we consume is oil. Oil is 50%, gas 25%, and they are used to heat buildings, run the industry, and move people and stuff around. 75% of the total. Electricity only represents 17 to 18 percent of the energy we consume in Belgium. So the whole game is to reduce the part of oil and gas, which will come with some increase of the electricity. But it is for a good reason. So if we like to continue to run the industry to learn, to live, we will need energy. We need to use a better energy. An energy we can produce on our soil to increase our sovereignty, and we can produce at industrial scale without emitting greenhouse gases. And that we can do with electricity. So I would say as far as AI helps us at Luminous accelerate the electrification to move away from imported fossil fuels that emit a lot of greenhouse gas, I would say overall the balance is extremely positive as long as it helps us electrify.

Speaker 2 17:11

And others are doing it, huh? Because I think China, everyone who's visiting China now and then, will see the massive uptake because not only the the cars that we are seeing driving here, but the economy itself. I don't know if you have any data points on that.

Speaker 1 17:24

Oh yes, uh China is well understood. Actually, China and Europe, in a way, they tell the same story. They tell the story that we need to electrify in order to increase sovereignty, to reduce the total consumption of energy and it reduce greenhouse gas emissions, and in China on top to increase air quality in the cities. These are the reasons why China electrifies. Now, Europe is telling the same story. Now, between us there is well a difference, is that China actually does it. So if you look

China Versus Europe On Electrification Speed

Speaker 1 17:55

at data, China is today at a share of its energy mix. If you look at the energy mix in China, 30% is electricity. In Europe, it is stuck at around 20%. And if you look zoom into Belgium, you're between 17 and 18 percent. We have the same strategy. What we need to do differently in Europe today is accelerate in the execution. So the reasons why Chinese do it, which is for a big part to reduce their dependency on import of oil and gas, we need we need, and these days, unfortunately, with what we see in the Gulf, um, confirm we have to reduce our dependency on oil and gas for the good of all the people in Europe and in Belgium. Okay, that's clear. If we bring it back to To get me warm, I need to take my jacket. That's okay. Or if I get engaged in the conversation.

Speaker 2 18:52

It's okay, it's cool. It's maybe the heating with electricity. Um but going back to the the the topic of AI as such. Um I think it's fair to say still that some people are still thinking it's a hype. Um, I'm convinced it's not uh hype definitely if you look at the core businesses of organizations itself. What would you basically say to people that are listening in and are still saying we're just gonna wait until everything is crystallized a bit because there's still a lot of trial and error, which is normal because we don't know what the end game of AI will be? Um what would you say to them? Uh and why would uh yeah, just wait and see until the others have tried it out and then jump on that train at a later stage? Why is that a bad strategy, quite here?

Speaker 1 19:44

You know, when I you you you work at PWC, no nobody's perfect. When I started my career, I was at uh Arthur Anderson who became Deloitte, and I remember very well uh 1995, I was a junior auditor, and I was making my work on, we called it uh 14 columns paper, yellow, double pages, and we were writing debit, credit, column one, column three with different colors of column three, the number of the account, and column four the name of the account, column five, the amount, and and so on and so on, and then some notes. So we were uh writing on four 14 columns paper, that's how we called it. And um only year two of my professional life I received the computer. Um I mean, what would have happened to Deloitte and PVC if in 1995 in Belgium you would have said

Why Waiting On AI Is Risky

Speaker 1 20:40

we will continue to do audit on paper because it might be safer than switching to the computer? I don't think it would have worked really well, and your customers would have been happy with the service they would have got. So today I see AI and large language model as a technological evolution. As any technological evolution, it's neither good nor bad, it's technological. But it's up to us to make it good. And at Luminus, we can use it to improve customer experience, we can use it to deal with increased complexity in the electricity system, we can use it to do better weather forecast, we can use it to integrate more renewable energies into the network, we can use it to reduce the bill of our customers. So these are all uses which we are doing today. Uh and where AI is our friend to help us do exactly that. Yeah. And I would say to those who hesitate, well, you know, probably just try and uh be curious, and you will be you will be flabbergasted.

Speaker 2 21:47

Because it's also because it's in fact not a tool discussion. I think it's there that probably some uh people are uh hesitant because they they say, Okay, will it now be tool X or Y that will eventually survive? But that's That's not the debate. Well then you pivote. Exactly.

Speaker 1 22:02

So we have just already pivoted. Once you have that so we we actually the core is uh I mean the our strategy has nothing to do with AI. We like to build a CO2-free energy future that didn't change. We like to do it by producing more electricity with a diversified mix that didn't change. Supplying energy, supplying energy efficiency services, provide flexibility solutions to the network. That's our strategy. It didn't change with AI. Now AI is bringing some vitamins to do that even better, in a more relevant way, with a more positive impact. That's more the way I see it.

Speaker 2 22:42

With a lot of down from the top five cents. You're driving it yourself pretty.

Speaker 1 22:48

Yes, yes. I mean, I'm really happy to be able to count on uh the critical support. I I mean initially it was mostly uh carried by our CIO, um uh Bruno Brusselmans, who who left us to continue with our colleagues at EDF Trading uh in the group uh in London, and he was the one to pull it. Uh these days, my colleague uh Henry Bunen, who is more on the uh on the operations, customer operations, sales, all the retail activities, is taking sponsorship. It also says something. So it means, of course, we keep a very strong anchoring into IT, but we also need the front office to pull and the front office to want to use AI into use case, into all sorts of processes to have this positive impact for our customers. So we have the sponsorship at the highest level. Obviously, uh my support, but also key people

CEO Role: Strategy And Safe Experimentation

Speaker 1 23:50

on the executive committee are pulling the effort at Luminous.

Speaker 2 23:53

And what's according to you the role of a CEO and all this?

Speaker 1 23:58

Well, I think the role of the CEO is probably to provide clarity on the strategy and then to make sure that we create the safe space where people can actually unleash uh the power of AI. I mean, I'm each day that goes, I'm flabbergasted by what our people do with it. So we provide uh environments, technological environments, some um ethical framework, some governance framework, but then we it's really our idea is to unleash and give authorization for our different businesses to experiment and when things are ready to actually launch. Um and and probably giving that authorization and and this I would say this energy is the small part I can do, and then write uh data um analysts, developers, uh colleagues who know processes, who know optimization, who know customer service, who know predictive maintenance of um power plants, then they do the miracles. It's mostly, I would say, it's mostly all my other colleagues who do the job. Yeah, but you are facilitating, I think that's important.

Speaker 2 25:17

I I I just encourage. Yeah. Maybe to come back on the data point, because I think it is something maybe people will

Data Quality Lessons From Energy Crisis

Speaker 2 25:25

not immediately make the link, but I think we're all sitting on vast amounts of data. So whatever sector you're in, AI will use data. So, and that's of course then the question of garbage in, garbage out. Yes, again, it's not new, but of course, AI puts an incredible accelerator uh on that. True. How are you dealing with that? Because of course you said it's part of your model, uh, it's the user experience will be increased by use of AI.

Speaker 1 25:53

So, how is I would say in every difficulty there is an opportunity, and um the biggest difficulty in my career was the energy crisis. So when was it? 2021, 2022, um, Russia invades uh Ukraine, and then suddenly uh Russian gas stopped to flow into Europe. It used to represent 40% of the inflow of gas in Europe. Gas price goes through the road. And um at that moment you don't know what happens in a week and two weeks. You just see every day what is the gas price, and at some point it reached 300 euro per megawatt hour, whilst before crisis it was 15, 20, so times 20. So price gas prices go times overnight. Yeah, so and then you're really there as an energy supplier, and you say, I mean, how do we manage our processes? How do we deal with our customers, exposures, issues? This was for me the biggest risk situation I faced in my business life. So Luminous has reacted at the time by uh strengthening processes, have a better uh risk approach, um, clean data, um uh let's say really make big processes and and improve data governance. We start, I mean, it was not new, but we tried to reach another league in quality during the energy crisis. And when we went out of the crisis, probably other processes, data governance, quality of our data was better than when we got in because we just faced risks at a level which had been unseen compared to our experience in the decades before. And now, when last language model came and uh AI, as we call it, tools arrived, we were working for several years with extra attention on all these dimensions processes, data, governance. And so it helped us reap benefits faster because at the end, AI and large language model is like tools that can help you get benefits faster, do things faster, program faster, uh, deal with the customer requests faster. But for that, you need a very solid data set, as you say, garbage in, garbage out. Now we would work, which is a very long haul work for years

The My Luminus App And AI Chat

Speaker 1 28:30

on the data and the processes, and so that really helped us go faster and reach benefits faster.

Speaker 2 28:36

Okay. Maybe uh a bit of a uh a string of shorter uh questions, so don't think too long, just uh respond, I think, on the on the go, I would say. What's the first AI tool or AI-enabled app that you use? My typical day, yes.

Speaker 1 29:00

Well, uh, you know, in my private life, I I I've tried uh Chat GPT uh obviously uh a lot. Um now we use at Luminous different large language models and that evolved. So we're talking about the choice of tools. That's really something for our IT department to find out which is the best tool to use. And for end users, not so interesting. What they care about is when they go in the My Luminous app, I mean, it's impressive the number of customer interaction we have in that app, it's really impressive. It's like 80% plus of all customer interaction goes through the app with a full treatment in the app. So people don't need to use another channel because they are dissatisfied. And uh in this app today, you have you don't have an old chatbot which was like a decision tree, and you could tell by interacting with it, like going square one and then square 1.1 and 1.2, and they're really rigid. That's all the world. Now we provide data sets into that system, and we can converse with customers on their own customer data set, and so we can converse with the customer with their personalized data with a natural language. So that that's how far it is today, and that makes a world of a change for customer interaction in an app, just to name one example.

Speaker 2 30:32

If you were 18 again and you knew then what you knew now, would you have studied something else?

Speaker 1 30:39

Well, you know, already at the time I was hesitating. What did you study by the commercial engineer at UC Louvain? And already at the time I hesitated between uh virtual engineer, civil engineer, and commercial engineer. And uh, you know, probably I was not brave enough to pass the entry exam into a civil engineer, and I studied commercial engineer, which I ended up loving,

What AI Changes Next And Closing

Speaker 1 31:05

and I I I really went along well with other students who still are my friends uh today, so no regret at all. But I hesitated between the two. I I think still uh that a solid uh education on um data, um uh mathematics uh is STEM. That's I would say is a plus uh for sure. If you can on top compound it with uh let's say a curious personality, interested, eager to learn and and and to uh agile, uh having fun, uh well that that probably is a great mix.

Speaker 2 31:48

Okay. How optimistic are you and how do you think that the AI will evolve, let's say, in the next four or five years, and also what it could bring Belgium as a country in a geopolitical context which is not that easy?

Speaker 1 32:04

Well, uh I I I think it just it's just like other major uh technological evolutions, like when you and I, I don't know about you, but when we started our career and we started working on paper, today it seems pretty obvious we do it on a computer. Now, if you you if you look at the speed of adoption of AI versus the speed of adoption of computer, it just is much faster to reach a critical mass. So, how it will evolve, I would say it will redefine the standards in a number of things like customer experience and it will set a new floor. So what people will expect will be an excellent, flawless customer service enabled by AI, but customers won't buy AI. They will buy relevant information, ease of use, fair price, uh, these kinds of things. So it will just redefine uh, I would say, customer expectations.

Speaker 2 33:02

Can are we innovative enough as a country with respect to AI? Is there the differenceator that we could use because we don't have a much of uh heavy?

Speaker 1 33:11

I would let companies do. Um I I feel we can do. I don't feel constrained um in Belgium. And I think um private initiatives probably will pull. So um I I I would let entrepreneurial initiatives uh be unleashed.

Speaker 2 33:30

Okay, with that optimistic tone. Thank you, Bigways. Thank you. Thank you very much. And thanks for again following an episode of AI Unscripted. More to come, so definitely stay tuned. Thank you, bye bye.