Episode Transcript
[00:00:00] Speaker A: Foreign.
[00:00:06] Speaker B: Hello everyone and welcome to this live podcast on AI and energy in Symbiosis. And my guest today is Stavros Stamos Tukos, Policy Officer Digitalization and AI in DG Energy Digitalization is one of the most important topics for the energy sector currently and there's further complexity with the fast evolving availability of AI companies and countries are developing many initiatives and at the same time the European Commission is working on a range of policy and legislative activities to guide and harmonize these activities.
And one of these will form the main focus of our conversation today, the Digitalization and AI Roadmap.
But before we start, I'd like to ask Stavros to sort of set the background with some of these initiatives that will really dictate to some extent the content of the roadmap.
[00:01:10] Speaker A: Thanks a lot Jonathan and thanks for having me on stage for this podcast today.
So let me give you an overview of the AI landscape in Europe and how the European Commission is working to address this upcoming technology.
So in 24 as you all know, the European Commission has adopted the AI Act. It's first of its kind regulation that takes a risk based approach on the AI applications and how they are deployed into different sectors of the economy.
If I can highlight two critical milestones of this AI act is the beginning of this year in February of 2025 that the enforcement period started for the prohibited use cases. So the sectors where AI is not allowed to be used in Europe and we're talking here about social scoring systems or real time biometric identification.
Next year, next summer we will have the enforcement period starting for the so called high risk use cases.
And this has implications to the energy sector but we can discuss about it later.
Then we have the AI Continent Action Plan. This is a strategy document that introduces the notion of the AI factories and the AI gigafactories.
This is a European funded initiative to create the sovereign cloud infrastructure where AI models in Europe will be trained and inferred deployed in different sectors of the economy.
Then we have the apply a strategy that defines concrete measures and flagship use cases per sector.
How AI can be deployed and energy is one of the sectors highlighted there in the strategy.
And I will close with an upcoming regulation is the the so called Cloud and AI Development act that will create the conditions for new cloud infrastructure. And I'm talking about commercial data centers. So it's not the AI factories that I mentioned before how data centers can be permitted and deployed in Europe in a more streamlined, accelerated fashion giving emphasis in security because these data centers potentially will be used by public authorities and also energy efficiency, which is directly linked to our sector.
[00:04:09] Speaker B: Could you go into a little more depth in how these will impact the energy sector?
[00:04:13] Speaker A: Definitely. So the main impact coming from the AI act is for the so called high risk use case.
So there is one of the high risk use cases is directly linked to the energy sector.
So AI, when it is used for safety of critical infrastructure, which is electricity networks, heat networks and gas networks, is considered a high risk use case.
In that case, the developers of the applications, but also the deployers. So the companies that operate this critical infrastructure should take certain steps to ensure that a risk management system is in place, a documentation system is in place, human oversight is in place.
So this might imply some extra work, some extra cost for the operators of the systems or. But this will ensure the security of our critical infrastructure.
The second of this action that will impact the energy sector is obviously the Cloud and AI Development act that aims to triple the data center capacity in Europe over the next five to seven years.
We are talking here about almost tripling the demand from the data centers, which might be not an easy task with the conditions we have in place today.
It's a huge demand that comes on our way and the energy sector has to adapt its processes, its regulations and the speed that can take this demand into the system, but not at the expense of other sectors of the economy.
Not rising the cost, not taking the demand from other sectors and not creating congestion that we already see in some member states already.
[00:06:19] Speaker B: And turning to the strategic roadmap on digitalization and AI, what are the aims of that and can you give a brief overview of the roadmap?
[00:06:31] Speaker A: Certainly. So the. The European Commission has been working on a digitalization action plan already since 2022. But back then it was a little bit before the.
The language models like ChatGPT came became a commodity in the hands of consumers and it was not in our either. So our, our Commissioner asked, asked us to develop a strategic roadmap for digitalization in AI in the energy sector. And we have two basic objectives.
One is to facilitate the development of trusted AI applications for the energy sector and their deployment securely into the energy system and on the other hand manage the risks. So we have to be very conscious that AI is a technology that moves very fast and we can easily lose sight and forget our main objective in the energy sector, which is to keep the system on and running safely.
So we should put a system in place that we can monitor the risks and act accordingly. So these are the main objectives of the Strategic roadmap and I can explain a bit more details during our discussion.
[00:07:53] Speaker B: And there was a call for evidence put out and what were the main points that came from that.
[00:08:01] Speaker A: So we launched during summer public consultation, which was published for three months. It closed early November. And just before coming here in Enlit, I got a preliminary analysis with which I can share with you today.
First of all, I have to say that we received a huge amount of contributions from stakeholders across energy and digital sectors. This is something very positive because this strategy brings both sectors together.
And if I can highlight one point that stood out in the consultation in the analysis is data sharing. So AI is going to be trained and deployed over access to data and data. Currently, Europe is fragmented, siloed behind either commercial systems or at member state level are not standardized and the access to it is extremely difficult for companies who want to develop smart energy applications, AI tools and deploy them.
So this was the main ask from our stakeholders to create a system, a governance framework where data can be easily accessible, but at the same time in a safe way, in a trusted manner.
Because we're talking very often about private data, data from our smart meters that have been, that need to be accessed by developers to develop AI applications. So this is something that stood out in the consultation and we will act on it. The second thing is the pace of development.
Many, many stakeholders pointed out that there are exceptional tools that they are being developed currently in Europe, but at very small scale. We are not able to scale our products into European or even better global markets.
And again, we want to create the conditions there to see what is needed by the market to scale these products in Europe, but also globally.
A third point that stood out was the energy consumption of data centers.
We can discuss later about it, but it's a new animal for the energy system. There are huge loads that they want to connect fast and the energy sector is an enabler to this process. So the energy sector has to be able to connect this load securely and safely. And this was a big ask. And then the stakeholders asked the Commission to create a level playing field across Europe. So a governance model so that we can monitor the digitalization of the energy system system across Europe over the next five years and monitor the progress, drive investments where it's needed. So these are the main messages that I can bring today.
[00:11:09] Speaker B: We'll come back to some of those points, but the common European energy data space would be a key enabler for the data. And what's the status of that?
[00:11:19] Speaker A: Definitely it will be so. The European Commission has funded Significant, significant investment has put a lot of money through Horizon Europe program in data spaces.
The whole program started in 2022. We have funded five projects that have delivered exceptional results. You can find most of them in the EU project zone here. Today most of the projects are about closing and there is a follow up project called NCM funded by Digital Europe program, which aims to deploy the common European energy data space in 16 member states.
So what we got out of this investment is the technical tools we need. So we have the technical implementation in our hands to go forward with a governance of the common European energy data space. So we have the tools to connect the federated data pools across companies or across member states, the so called connectors. We have a trust framework, an identity management framework. So all these are tools that were developed by the projects. What we have been asked in the public consultation is to create a European layer where all these will come together and will allow companies or companies countries to exchange data, but in a federated way. So this is for training AI models for example, or for aggregators to provide flexibility services.
So the common European energy data space, it will be the basis of the governance layer that we have been asked to develop in Europe.
[00:13:09] Speaker B: Another aspect is research and innovation funding. And is there a prospect of more funding from the European Commission?
[00:13:17] Speaker A: Definitely. So as I said, we have invested a lot already on data exchanges.
The new call For Horizon Europe 2627 comes out in a few weeks from today, but drafts are already available for you to see. So there is a lot of funding, not just for data spaces, but also for modernizing and digitalizing our energy system system. I can give some indication we will be investing more than 80 million euros in smart control rooms for transmission system, for distribution management systems, but also more than 50 million Europe for development of AI tools. So we have dedicated calls for data exchange, for training AI models, we have dedicated calls for creating AI monitoring tools for grid operations, but also AI control systems for grid scale storage systems.
So this is coming already now, but also we are looking in the future. So we have proposed a significantly increased budget under the new MFF which is currently being negotiated with a code legislator in different categories. So in Horizon Europe for research and innovation, but also in CEF Energy, this is the fund that finances infrastructure.
There is a proposal for a fivefold increase there and we hope that we can see more smart grids and smart system projects funded by the safenergy. So yeah, that's, that's right to say that funding is necessary, public funding, but we are also working with Private investors to see how this public funding can leverage private, because public is not enough.
[00:15:16] Speaker B: Right. And you mentioned data centers. And what is the Commission's approach, policy approach to data center development?
[00:15:26] Speaker A: So we are, the Commission is working on the Cloud and AI Development Act. The objective there is to triple the data center capacity in the next five to seven years.
The stripping capacity for the energy sector. It creates a huge challenge, but also an opportunity.
The data centers are huge loads. So we are talking about gigabat scale. At least this is what some investment plans are showing.
And these are loads that are not common for the energy sector. So it's not common that a grid can connect a city scale demand in one connection point.
The pace of development is very fast.
So a data center can be built in two to three years, while a new grid infrastructure might need eight to 10 years.
So the challenge is huge. But we are working with the data center sector closely.
So the European Commission has tried started a proactive dialogue bringing together the energy sector and the data center sector to identify the bottlenecks. And I can reassure you that both sectors have been working together very closely. They have a common understanding of the challenge.
And we are organizing a series of workshops. The next one is in December in Brussels.
And this is only the beginning. We see the problem coming up, but we want to not focus on the problem, but on the opportunity. So data centers have an immense potential for flexibility. The International Energy Agency has quantified it to 50 to 60 gigawatt over the next five years. This is a lot. If we can tap to a fraction of that, this will be really beneficial for the energy sector. So again, the Commission is working together with both sectors and we want to structure this dialogue and bring it forward into a possible commitment either at European or national level, where the decisions are made.
[00:18:00] Speaker B: And will that feature in the roadmap?
[00:18:03] Speaker A: This will be featured in the strategic roadmap. Yes, this will be one of our pillars of actions and we will propose some early measures there, but the most important of which is to continue the structured dialogue.
[00:18:19] Speaker B: And what about the energy efficiency aspects of data centers?
[00:18:23] Speaker A: This is very important, the energy efficiency aspect. And I can give you a little bit of update. You know that in the Energy Efficiency Directive that was adopted by the Commission a couple of years ago, there is an article about energy efficiency of data centers. As a first measure, we proposed a reporting scheme. So data centers across Europe, above a certain IT load, I think it's 500 kilowatt, they're asked to to report on basic indicators, most of which is about energy consumption, but also water consumption, waste, heat utilization.
This is the first step to know how many data centers we have in Europe, where they are, how much they consume.
We are publishing reports based on the data that we receive. We collect this data through the member states and we published anonymized reports per member states on this energy consumption. This was the first step based on the reporting scheme that is in place now. So we have received data from almost half of European data centers already. And we are only one year into operation of the scheme.
Based on the scheme, we will bring forward early next year, together with a strategic roadmap on AI, a regulation for energy rating of data center. So based on the indicators that we receive from the data centers, data centers will be assigned a certain efficiency indicator.
It's not decided yet what whether it will be a lettering scheme or a scoring system. But we will know not only the consumption but also the efficiency of the data centers.
And again early next year, together with our strategic roadmap, we will launch a consultation with the sectors. I have to tell you that all these schemes have been developed in very close cooperation with a sector, the digital and energy sector.
And similar cooperation will start early next year on a possible new regulation on minimum performance standards for data centers.
[00:20:44] Speaker B: And how will the energy sector tap into the AI factories and gigafactories?
[00:20:50] Speaker A: So it's a huge opportunity for the energy sector. There are at least three AI factories that will have an energy vertical in their scope.
So we are working with all three of them.
And the objective is to leverage the compute capacity and sovereign infrastructure of these AI factories. They're all based in Europe and they're using, in most cases, public cloud infrastructure.
So we are leveraging the compute capacity of these data centers of these AI factories to train models for the energy sector.
This is a very early process, so we are just starting now. And I invite all stakeholders who are interested to be part of this collaboration to join the discussion.
So the objective, as I said, is to to jointly train AI models, test them and validate them in regulatory sandboxing environments and then safely deploy them in the energy sector.
[00:22:12] Speaker B: And the important thing with any legislation is monitoring the implementation and what sort of plans are going to be put in place for that for the roadmap.
[00:22:24] Speaker A: So the objective is to work with member states and set clear targets for digitalization or for AI.
This is early stages.
We don't have concrete ideas how to go about it, but we indeed want to create a monitoring framework and a governance framework around it.
So we want to work closely with the member states to see how we can decide on common indicators to monitor the digitalization of the energy system and then meet regularly and update our plans. As you know, AI and digital technologies is a really fast paced environment, and the speed that we are used to working at public policy level is not fit to follow these developments. So we need an agile structure to update our objectives, to update our indicators often and keep up with the technology and the new challenges it poses for the energy system. This is very important, the agility and the flexibility.
[00:23:34] Speaker B: Okay, good. Well, thank you very much, Stavros.
[00:23:37] Speaker A: Thank you, Jonathan.
[00:23:38] Speaker B: Thank you very much.
