Mr. Chairman, Mr. Chief Executive Officer, good morning. The Committee on Economic Affairs is pleased to welcome the founder and Chief Executive Officer of OVHcloud, Mr. Octave Klaba. We thank you for being here and welcome you to our committee. We wanted to organize this hearing to discuss with you the development and strategy of OVH, of course, but also the development of data centers and cloud computing, as well as the issues of data sovereignty, transparency of services, and the development in Europe and France of an innovative and independent digital ecosystem. In general, our committee is very attentive to the development of our economic and productive capacities, as well as to the dynamism of companies in the digital sector, their competitiveness, and their contribution to our industrial sovereignty. OVHcloud is today the leading French and European cloud player. You generated nearly 1.1 billion euros in revenue last year, and your group has more than 1.6 million customers and 46 data centers spread across 9 countries. Your group, which invests about a third of its revenue, is therefore an essential player for our country in the strategic field of data centers and cloud computing. I would also note that your group obtained, on September 1st last, the SecNumCloud 3.2 qualification granted by the National Agency for Information Systems Security (ANSSI) for its new public cloud platform, SNC Cloud Platform, which complements the two qualified offerings your group already has with BareMetal Pod and VPS. I recall that SecNumCloud is a qualification issued by ANSSI that certifies the security and sovereignty of a cloud service offering. Your group was also selected, along with Dassault Systèmes and Clever Cloud, for the sovereign cloud market of European institutions, which represents 18 million euros over 6 years. However, OVHcloud, despite its importance and successes, remains significantly smaller than the major American players in the field, what we call the hyperscalers, that is, the cloud computing service providers, in plain French, the storage and data management players who operate very large-scale data center infrastructures. You stated on April 21st last, before the inquiry committee of our Assembly on structural dependencies and systemic vulnerabilities in the digital sector and the risks to France's independence, chaired by Mr. Lombrail and whose rapporteur is Ms. Cyrielle Chatelain, that European public procurement was still counted in millions, while building a cloud champion required a scale of billions of euros. And we know that our companies, in order to ensure their development and remain competitive, are increasingly required to use advanced digital services, including artificial intelligence, which raises the question of economic dependence on non-European companies and the protection of our sovereignty in data management. Before giving you the floor for about fifteen minutes, allow me to ask you a few initial questions. What trajectory can you envision for the future of OVHcloud to change scale by 2030, and what is today the main obstacle to this development? Your group has invested several hundred million euros. How do you finance a much stronger acceleration while making it sustainable? You stated in April before the inquiry committee I mentioned that European public procurement remained very insufficient to foster champions with the rules of public procurement. What rules of public procurement do you think are necessary to strengthen it? The sovereign cloud contract of European institutions, which could reach 180 billion euros over 6 years—million, sorry. Do you consider this a real change of scale or rather a market still too limited given the necessary investments? And your new SNC Cloud Platform has just obtained the SecNumCloud qualification. What are the next steps, and do you anticipate any difficulties? You have 46 data centers spread across 9 countries for administrations and critical operators. What service continuity guarantees can you provide today in case of serious problems at one of these data centers? The location of data centers also raises questions of access to electricity, land, and water, among others. Do you identify obstacles to your investments in France on this subject? And finally, what recommendations would you make to us legislators here to reduce our dependence on non-European clouds in the medium and long term? Thank you in advance, Mr. Chairman, Mr. Chief Executive Officer, for the answers you can provide us and your participation today before the members of the Committee on Economic Affairs. I give you the floor for about fifteen minutes.
Good morning. Thank you, Mr. Chairman. Ladies and gentlemen, Members of Parliament, I thank you for this invitation to speak before your committee, which brings together industry, energy, innovation, and digital—a subject at the very heart of OVHcloud's activities. Mr. Chairman, you recalled our successes but also the gap that still separates our activity from that of the hyperscalers. This gap is real. We will address this question. They all come back to one thing: how to change scale? Critical mass—I will come back to that after a few words of introduction about OVH. OVHcloud is 27 years old. I founded OVHcloud in 1999 in Roubaix, without capital. For more than 16, 17 years, I grew the company, always with its own capital. The first fundraising I did was in 2016, 250 million. The second fundraising, so to speak, was the stock market listing in 2021. We also did a capital reduction where I returned 350 million to investors, and so the company evolved with little external funding. Yet today, we generate more than 1 billion in revenue. We have 3,500 employees. We are deployed in Europe, the United States, and also in Asia. But it's a story of freedom and independence. Probably because my personal history led me to be a little bit paranoid about how to build companies and to see things in the very long term. When I started OVH, I already had 20 years ahead in my mind, and today that's still the case. Today, OVHcloud meets all cloud needs and soon more and more needs that we call artificial intelligence today. I mentioned our revenue; it's also important to see that we serve different types of customers. We of course have the public sector in France, but not only. In Germany, in England, what we call Rest of Europe, meaning England, the Netherlands, Poland. We are also in Canada, we are also in the United States. So we have these business sectors. But if we had focused solely on this type of activity, we wouldn't have had the critical mass necessary to build a company of this size. And that's why we come from digital, let's say, where customers can simply come to the website and order services with a simple bank card. We also have that activity, which is very important since it represents today about 60-70% of our revenue. The corporate, private and public companies, represent about 20% of our revenue. And so it's really very important to understand that in our business, critical mass, critical size, is essential to build this type of company because the investments behind it are so important that if you only go after one type of customer, you won't have a return on investment and you won't be able to continue making investments every month, every year, and continue to grow the company. Our ambition is to reach 2 billion in revenue in the coming years. I won't give the dates yet because we're a listed company and, as you know, it's very regulated and investors scrutinize these types of investments. I will answer the questions you asked me, and perhaps on the first question I will answer at the end. What changes to the rules in public procurement? I was saying it's really about critical size, meaning that today, if you take for example OVHcloud in France, the public cloud of the State is 84 million euros annually in 2025, with OVH as the number one with 33 million. 33 million, put that against 1.2 billion in revenue. So you see, on this, it's great, there's revenue, but it's rather in the tens of millions of revenue. And so compared to many of our competitors who are unfortunately more modest in size, this type of revenue in the tens of millions. Well, for a growing player heading toward a billion in revenue, who is European, who is global, who has a presence in many countries, who has many clients in 140 countries, but in fact this public procurement of around hundreds of millions or tens of millions is anecdotal compared to the overall growth needed for a cloud and artificial intelligence company. But it's good, it's good. Things have changed in recent years, and we've seen these progressive earmarks of public procurement to use European infrastructures. We host, for example, AIFE, which is the electronic invoicing system implemented in France on September 1st. So at AIFE, we host many intermediary service providers, and so all these elements are part of our clients. But these are quite modest details today compared to more global growth. The cloud contract of the European Commission. It's very interesting because following the various shocks we've experienced in the last 2-3 years, the American administration is one of them, but it's not the only one. There's another one called VMware that was acquired by Broadcom. All these shocks that have arrived recently but in a somewhat unpredictable and unforeseeable way have created reactions within private or public companies. There are two types of reactions. The first reaction is: well, it will pass, and at some point things will return to normal and it will be fine. So we keep a low profile and eventually all this will settle down. And then there are others who say no, we really need to change our approach, and the European Commission is one of them. They created a framework. They used these frameworks to select European players. We were chosen as one of the winners, along with Dassault Systèmes and Clever Cloud, on a contract of 180 million over 5 years. That's good. That's good because it's the first step. I think it's the method that matters more than the amounts at stake today. It's true that 180 million over 5 years across four players isn't everything that needs to be done. But it's the method that interested me more. That is to say, the European Commission says: look, we're going to put 180 million to test, to develop with you, to give you the possibility to go after tenders from the various agencies and to prove that you're capable of taking over all the activities or at least part of the activities that are today hosted on hyperscalers. I think that's super interesting because you have to put these 180 million in perspective, which globally represent about 40 million per year across 4 winners. On the other side, you have about 1.2 billion in revenue that AWS generates with the European Commission per year. Per year, okay? And you have another 500 million with Azure. And so you have to put that in perspective. That is to say, you have 1.72 billion in revenue generated between the European Commission and the hyperscalers compared to a first tender. So you have to see the method, that is to say, these are big businesses but also a lot of responsibility. It's a lot of products, it's a lot of things where you need proof points. What we call proof in English is being able to prove that you're a trusted player. And so it's the method that interested me most. If you look more broadly at Europe, there's about 10 billion in revenue generated between the various governments, the various commissions, the various ministries with the hyperscalers in Europe. So the critical mass isn't there. It's not about creating new budgets. It's about redirecting them toward European players and being able to trust them to help them grow. And so the money is already there; it's just missing a little bit of courage to make sure that money goes to European players. Another question you asked me about the SNC and the other platform we just obtained. What are the next steps and difficulties? I like what ANSSI did because they brought a technical framework, unlike all the other standards which are just norms saying figure it out however you want. They went much further, much further, saying how it needs to be done, bringing a certain know-how they have in securing infrastructure. And that's what makes it extremely complicated and long to obtain because, well, it's not just about saying I have papers, I have processes to build the infrastructure. And so we fully embrace these frameworks. We're already used to it; this is the third certification we've done with them, and we'll continue to enrich our products. Today on the cloud platform, we have a number of products we're going to enrich. The goal is for this portfolio to be as rich as possible and to benefit all clients, including the State and private players. We have 46 data centers in 9 countries. What guarantee of service continuity? It comes from the technical design we do around the data centers but also around all the activities that take place in these data centers. The operating principles: having a single data center is in the past, that's the 2000s. Since 2020 and in the coming years, it's rather distributed data center services. That is to say, you have what we call a region of three data centers spaced 30 km apart, powered by different energy sources with different fiber optics, so that everything the clients run can also tolerate a failure without impacting overall operation. And so it's these types of increasingly distributed designs that we're deploying and changing the entire configuration of our data centers. You have the same types of principles in AI inference. That is to say, today we see very clearly that inference—so that's everything that allows you when you type in the chat to receive tokens—we see the criticality of these types of activities. So it can't fail. Okay. It's also important. It becomes as important as, for example, telephony, as all communications, because there are many companies that depend on tokens, and so you can't put all activities in a single data center in a single location. You have to put them in many different places and connect them across the network through different schemes. In any case, that's our approach to providing resilience. Electricity, land, water, obstacles to investment in France. Well, there's a long list of regulations. I started 27 years ago. Okay. So I have a little bit of a track record on managing the different issues in different countries. It's become more complex in recent years to the point where you count in several years to deploy a data center or even upgrade it. You just have to look at ICPE: when you have a data center and want to expand it, you must of course file the application, that's normal. Okay? But the fact of filing applications blocks all investments on sites for 2 years. That means you're prohibited for 2 years from doing anything on the sites. So I challenge you to manage a business with this kind of growth where you have to grow quite rapidly and you're told: don't touch anything for 2 years at that location because we're studying your application and we'll tell you in 2 years if you can grow. You can't. Okay? You can't. And the list of these types of constraints goes on. I'll give you an example at our Roubaix site. So we're from Roubaix, we progressively bought different buildings in one neighborhood. Okay? So each time it's a different address. The buildings are 200 meters, 500 meters, sometimes 700 meters apart. And then progressively, we filed ICPE applications at the different locations. But since these were sites of 2 MW, 5 MW, it went relatively well. But at one point, they came to us saying: you must consider all of this as one site. Okay? And so from now on, we want everything to be considered as if it were one address, the whole neighborhood. Okay? So we were blocked for a certain time to bring all this into compliance, so to speak. Rules were imposed on us saying: well, you have different buildings, different locations, it's not a big deal, you must consider them as a whole. So you have a certain number of investments you must make for a certain number of constraints. We had to spend, I think, 5-6 million additional to bring a whole neighborhood into compliance, so to speak, with the complexities of distances and all that. That's just one example. Okay, we live with this, and it's a bit longer than without all these constraints which are important in the abstract, they protect us, they ensure we operate in a safe place, but still, that's the reality for entrepreneurs, that's the reality of this business. And today when we talk about data centers, I don't think we'll have gigafactories in France or even in Europe. That is to say, places where there will be 1 GW of electrical power deployed in a data center at one address—for me, that will never exist in Europe and in France. It's just that the Americans are building with all the feedback we're starting to get, 2 GW or 3 GW with all the issues around CO2, energy, off-grid, water consumption, etc. Well, in France that will never happen because it will simply be blocked before. If you want to build a 100 MW, 200 MW data center today, you already need 4 to 5 years to get there, even with everything that was done previously by the President to push this whole activity. And what are the recommendations for legislators? I like freedom of action, and that's what I like: that personal initiative, individual initiative can change things. That someone gets up in the morning, has an idea, takes risks, creates businesses. And so I like to let companies act and give them as much freedom as possible. But I also realize that we live in a country that's a little bit undisciplined, that isn't focused on the long term, that's satisfied with a certain pace of operation or activity from one year to the next, and we don't really like change. So I just have one idea, and I don't really like that it be done by law, but it's really a reflection I leave you with: it's the 15% of activities. That is to say, you don't need to redirect 30, 40, 50% or earmark 50% of digital activities toward European players. If you put 5, 10, or 15%, you'll create giants in Europe, in France, in digital. It's enough because it's a certain critical mass that's already not so small, which allows behind it to develop, to invest, to rotate capital, and to ensure we have more of these types of companies in Europe and in France. And the second element, what I'd say, but I don't know how you can intervene on this, is financial culture. Financial culture in Europe, in France, is very much oriented toward: I invest, I lose money. Okay? So it's somewhere: are you going to invest money? Are you capable of losing money? Profitable investment, making investments that then create value or drive growth. We're in this start-up nation approach. Somewhere we've lost sight of that. I created the company without outside money, and that's my DNA. I don't know how to do it any other way. And that's what we don't have today. On one side we have investors who want cash, free cash flow, dividends, and on the other side we have a whole population that says: well, I'm willing to raise money not to make a profit, to do developments, to do things that amuse me, but I'm not really looking for revenue or activity or anything sustainable. Fortunately, between the two, there's still a whole range of companies that are on the right topics but details insufficient at the moment.
I propose that we stop here for the opening remarks, Mr. Klaba. We'll give the floor to the group spokespersons, and on the topics you'd like to come back to, obviously either through questions or at the end of the hearing, I propose you do so if that suits you. That's perfect. Thank you very much for these presentation elements and the first answers to the questions you were kind enough to give us for the Rassemblement National group. Mr. Édouard Jordan, you have the floor. Mr. Member of Parliament.
Thank you, Mr. Chairman. Mr. Chairman, Mr. Chief Executive Officer, your hearing is an opportunity to highlight the paradox of French digital. We have the skills, technologies, and industrial players across almost the entire sector value chain. Yet our administrations still largely depend on foreign suppliers. The government is slow to fully grasp the issues of digital sovereignty. The example of the health data platform illustrates these contradictions. By entrusting health data hosting to Microsoft, the Macron government created a breach regarding the extraterritoriality of American law. Your company had positioned itself to take back this hosting. In April last, it was finally the French company Scalway that was chosen to begin the migration toward a sovereign cloud. How many years were lost before this change of course, when French players had the capacity to respond to the tender within acceptable timelines? This lack of state support is all the more concerning given the threats that foreign competition poses to our digital industry. The takeover of Atos, the historic champion of French digital, reminds us that industrial sovereignty can be lost for lack of political will. We cannot ask our companies to invest in tomorrow's technologies while leaving them at the mercy of foreign funds. French money must support our industrial capabilities and ensure the security of data entrusted to administrations. For the Rassemblement National group, this assessment calls for a clear response: establishing national priority in public procurement. On October 8th, we will carry in our parliamentary niche the resolution introduced by my colleague here, MP Alexandre Loubet, which calls for allowing public buyers to favor offers that create jobs, production, and innovation in France. We thus want to favor French players like OVH, Scalway, or Clever Cloud in the State's digital projects, associating our companies with competitive nuclear energy. We can rediscover an industrial ambition equal to the era of data centers and AI. Mr. CEO, would you be in favor of establishing national priority in public procurement? I thank you for your response.
Thank you, Mr. Member of Parliament. Mr. Chief Executive Officer, a few minutes to respond.
As I was saying, it's 15% of redirection of revenue, in any case of procurement, that allows creating European players of good size. You also have to take into account the critical mass issue, meaning France alone cannot today and cannot respond to this problem. The response must be European. If you don't have the European critical mass to feed revenue to French or European companies, you'll have champions as they are today on home soil around 200, 300 million, 500 million in revenue, because that's what cloud companies will generate if they focus solely on France. Behind that, you can create champions with several billion in revenue because you'll have access to multiple markets. And so why does OVH today do 1.2 billion and we're on the path to 2 billion in revenue while all our competitors are rather around 100, 200 million? Because from 2004, after 5 years of company growth, I decided to deploy in all European countries, and I did that for 4 years. So I did Poland, I did Spain, Italy, Portugal, Ireland, England, the Netherlands, the Baltic States, Germany, Switzerland. So I did a lot of European development from 2004 to 2009. In 2010, we were asking ourselves: how do we get to the United States? We didn't go right away, we went to Canada, and so in 2011-2012 we did an expansion also through Canada and through various data centers in France and Europe. In 2016, for me it was the last moment in our history to say we need to become a global player. And that's why I did the 250 million fundraising in 2016 to build data centers all over the world. That's when I went to the United States. I did Singapore, I did Mumbai, I did Australia and other European countries like Poland, Germany, etc. That's the critical mass. And so the issues, the investments are so heavy that if you only have one country to count on, that's great, and for me you need to redirect part of the revenue to Europeans, but if France alone doesn't have this activity and this initiative isn't European, we won't be able to build companies worth several billion.
Thank you, Mr. Chairman, Mr. Chief Executive Officer. For the Ensemble pour la République group, Ms. Danielle Boullois. You have the floor, Madam Member of Parliament.
Mr. Chairman, Mr. Chief Executive Officer, on behalf of the Ensemble pour la République members of Parliament, I first want to salute your journey and your commitment. You founded in 1999 in Roubaix a company that has become the leading European provider of cloud computing services, present in many countries and employing several thousand employees. It's a French success story capable of competing with the world's giants, and we're very proud of it. This success takes on particular significance today. Digital sovereignty has become a leading issue for our economy, for our communities, for our hospitals as well as for our companies. Our data is a strategic asset, and it's imperative that public and private players care about hosting, protection, and the foreign legislation to which they may be subject. The SREN law aimed at securing and regulating the digital space and the SecNumCloud qualification go in this direction. It sets a demanding framework of which you are one of the reference players. In the Jura, as in all rural territories, our artisans, SMEs, agricultural operations, and municipalities are digitizing their activities. They're worried, for example, about electronic invoicing, which exposes our economy since one can know who sells what and who buys what. According to you, is there a risk? They need reliable, accessible solutions that protect their data. You operate in a market where American giants occupy most of the space. How do you analyze the current balance of power with these players? And what measures do you expect from public authorities? You've already answered a bit. Can public procurement be, according to you, an even more effective lever to support your sector? Facing the American giant, isn't the more modest size of a player like OVHcloud an asset in agility, proximity, trust, and cost control? Where are you in your investment strategy in data centers and artificial intelligence? Do you have projects for the coming years? And can our rural territories find their place there? Finally, how can we help and support you in training the talents you say are the lifeblood of your sector and which it needs, including in our territories far from the major metropolitan areas? I thank you.
Thank you, Madam Member of Parliament. Mr. Chairman, Mr. Chief Executive Officer.
Thank you for the many questions. I'll try to answer some; I won't get to all of them. On artificial intelligence, for me it's essential. For me it's essential and it's a real challenge. Today the technologies are mainly American or Chinese. Okay? And the innovations come enormously from China. In Europe we don't have technological players who want to invest in artificial intelligence, in the technology of artificial intelligence. Of course you have uses and how to implement them in companies, how to use them daily in processes and increase companies' productivity. But there's the technology underneath which is extremely important. Today many people place a lot of trust in open source, in DeepSeek, because it's good but it's Chinese. That is to say we don't know what will happen tomorrow regarding this source and we don't know what strategy lies behind it. What will companies adopt in a year, 2 years, or 3 years? Is this open source temporary to make us dependent on this type of model, or is it permanent? Will there be companies that want to invest billions and then always source the models? In any case, we came to the conclusion that it's too risky not to invest in artificial intelligence. That's why since the beginning of the year we've started making acquisitions of companies with talent to strengthen ourselves technologically inside the company, to then train models from scratch from the beginning and enrich them for the business areas where we need them. And we have three. The first area is around offensive or defensive security. That is to say we can see clearly how the security stakes around artificial intelligence have become colossal. These are real problems because AI allows discovering security flaws or penetrating systems at a speed never seen before. So at our level, we estimate we have a lot of our own data—not our clients' data—because we manage large data centers, we manage enormous activity, we suffer a lot of attacks, we have enormous amounts of data we can use to train our systems. We come from there, okay? We have to be good at security. Second element is coding, developing software. We use LLMs to develop, but we also want to have our own LLM models that will develop software. And third is agentic AI for the areas we have: finance, legal, HR, supply chain, and all these elements far from having money. And so we think that if we develop these LLM models for our own needs and we're happy with them, we'll also make them available to our clients so they can benefit. We could talk much more about artificial intelligence, but this is a very quick response.
Thank you, Mr. Chief Executive Officer. For the La France Insoumise group, Mr. René Pilato, you have the floor, Mr. Member of Parliament.
Thank you, Mr. Chairman. Mr. Chairman, Mr. Chief Executive Officer, Europe pays 264 billion per year to the United States for cloud and software services used. For France it's 54 billion, nearly 2% of its GDP. Several ministries still use the Microsoft ecosystem and the DGSI even extended a contract with Palantir at the end of 2025. Not to mention AI, France and the European Union depend nearly 80% on American digital giants for IT storage. According to the Future of Technology Institute, the security systems of three-quarters of European countries rest on these technologies. The United States could therefore from one day to the next restrict or block access to our own data. This makes France a quasi-digital colony and it's unacceptable. Faced with this dependence, the EU has chosen words rather than action. Talking about sovereignty when only 1% of the most sensitive uses are concerned is suicidal. Sovereign cloud is today too often an American offer with a European wrapper. It's a data center on European soil but the ownership, technology, and applicable law remain American. This allows the authorities of that country to demand access to data regardless of where it's stored. Europe arrived here because it gave in on its immunity requirements against extraterritorial laws under pressure from certain member states and American giants. Our markets continue to choose Microsoft, Amazon, or Google for lack of clear rules imposing preference for European solutions, while the United States and China protect their champions without complex. European offers for AI storage exist. What's missing is public and private demand giving them the necessary scale and above all rules that stop treating equally an actor subject to European law and an actor subject to American law. Digitally decolonizing our country is possible. OVHcloud is part of the solution. Mr. Klaba, you built a European world-class player against giants backed by a protected market. Within the framework of a major plan by La France Insoumise for digital decolonization, whose objective includes having French data centers. Three questions. Should we bet on mega data centers or on a network of small data centers to ensure our resilience? You partly answered in your opening remarks. Over what timeframe is OVH capable of responding to an industrial demand of this magnitude? How to increase the energy efficiency of data centers, reduce their water consumption, and recover their heat production? Would Switzerland with Infomaniak be an example? I thank you.
Thank you, Mr. Member of Parliament. Mr. Chairman, Mr. Chief Executive Officer, I'll start with the last question on ecology and data centers. It's one of my favorite subjects because since 2004 we've been cooling data centers with water. Okay? We don't have air conditioning in our data centers. We have 500,000 physical servers, so physical computers, that are cooled with water. So what does that allow us? It allows us to reduce our electricity consumption by half. So it's not anecdotal, it's a lot. And since all our accounts are public, well I can say today we pay about 70 million per year for energy across our various data centers. If we didn't have these technologies, we would have paid 140. Okay? So these are real savings we make, and it allows us to be cheaper on the market and offer cheaper services. So we pass this added value on to our clients. These are types of technologies we developed more than 20 years ago and have improved over time. The same types of innovations we made on the circularity of parts. That is to say we build all our servers in our own factories. We have two factories. We have more than a hundred people, 120 people, who build our servers. We don't buy servers elsewhere. We have the whole global supply chain where all the parts arrive in the north of France and near Montbonnoud, where we have two factories in which we manufacture a little more than 50,000-600 physical servers per year, which allow us to customize, to really reduce costs, and to put...
servers that are specifically good for the different jobs we have to deliver. True for our customers. And so in that way too, once the servers go into the data centers, after a certain number of years, well, in fact, you have to remove it, you have to recycle it, so the whole set of all this recycling chain we have put in place for years. So the whole set of data centers, we build a lot of data centers that we built on brownfield sites to reduce CO2 consumption. And there again, we can talk about it for about thirty minutes. It's a subject that was extremely important to us, to be able to reduce both the impacts and the consumption, and behind that to be able to deliver services at large scale.
Thank you, Mr. Chairman and CEO, for the ecological and social group. Madame Criel Châelin, you have the floor, Madam Chair.
Pardon.
Thank you, Mr. Chairman, and thank you for this hearing and for your remarks. I would have two questions. The first, you mentioned it in the answer about the supply chain you have built. Could you elaborate a bit more on the advantages of this supply chain and perhaps is it something that should be further developed to guarantee the sovereignty issues? And the second is about AI. You started to mention both what you develop internally and it seems to me that you also have an offer and deploy. Does that change things in terms of infrastructure management, in terms of energy consumption, or globally how do you approach the development of artificial intelligence in terms of managing your infrastructures?
Thank you, Madam Chair. Go ahead, sir.
On the supply chain, very good, well, it's a crucial subject. I'll just give you a figure, in fact right now we're going through a crisis in terms of the cost of memory sticks. Memory is the memory that you have in computers. You have that in cars, you have it in smartphones. The cost of the memory stick, in fact, of memory has increased by 6 to 9 times in the space of, let's say, 12 months. Okay? So it's a major impact on the whole set of cloud and artificial intelligence costs. And it continues to increase for a fairly simple reason, which is that a lot of global production has turned to artificial intelligence and the global production of the three suppliers has been reoriented towards GPUs, towards artificial intelligence and has neglected all the other activities on the side. And so, as a result, supply and demand, demand being always there. Supply has raised prices because there is no more global production. Europe, we buy 3% of the volume of global memory sticks.
Okay? 3%. Europe, the whole set of European territories, buys 3% of the RAM volume. That is to say that it's not here that it happens. Production, the whole set of assemblies, it's there, it's not here that it happens in fact. The putting in place of all these electronic activities. It happens in the United States and it happens in Asia. So Europe, it's been years since it's no longer up to date, well, it's no longer up to date in terms of electronics. We have, with Bull, we are almost the only European manufacturers of computers. There are not many others. There are a few factories in Turkey for HPE but there are extremely few. It's no longer made in Europe. There are very few, in fact we are dependent on all these activities that are outside of Europe. I'm not even talking about chips etc. or even the United States, they are having trouble bringing all this activity back to their soil. On artificial intelligence, there is a lot, a lot to answer. I think that I... so it's on AI, regarding the questions you had, innovation in artificial intelligence and the development of new models, well, it requires having GPU clusters in our data centers. Well, it turns out that in fact I made an investment in quite a few data centers about ten years ago, about fifteen years ago. I have 250 MW of electrical power. So about more than half used, but the other half now we will be able to accelerate their use. And so I'm a bit ahead in fact compared to all the phases of all the actors who say well I don't have more electrical power, I don't have more space and so I'm a bit less comfortable today in evolving artificial intelligence. Nevertheless, the technological designs of the data centers that host artificial intelligence are totally different from other types of infrastructure. The concentration of energy is enormous per square meter and that requires a certain number of electrical cables or completely different electrical systems. And that's where history proves us right again. Water cooling comes to save the cooling since you can't put so much electrical power in a square meter without being able to extract so much heat without glycolated water or water. And so being an expert in this technology for more than 20 years, well, we can see very well that we master this whole supply chain of GPUs and we have deployed at an extremely important speed in our data centers. But it changes a lot on the site, on the way of bringing and behind that you have to take into account that the technologies are still evolving. That is to say that if you invest today in a data center and you say "That's it, I have an AI data center", be sure that in 18 months, 12 months, you will have to do retrofit in your data center to be able to adapt it to new GPU technologies that will come out in 2 years because your energy or cooling chain will no longer be up to date. And so that changes completely the way of investing in data centers. You can't build a data center today and say that's it, I've built something for 20 years. You can build shells in which over time you will deploy technologies that are incompatible with the GPUs of the moment.
