NOTE: I regularly speak to you about artificial intelligence, its advances, and its applications. Today, I wish to address a related but equally important topic: the cloud! If AI represents the visible part of our digital revolution, the cloud constitutes its invisible foundations (and not just for AI). In what follows, a small uncompromising analysis of our technological dependence highlighting the growing danger it poses to our sovereignty. The choices we make today (or that we neglect to make) will determine for decades to come our ability to remain masters of our digital destiny. More than just an assessment, this article is a call to action.
Brussels, April 2030 — Europe is celebrating today the fifth anniversary of its “Cloud Digital Euro,” now considered one of the continent’s greatest technological and political achievements since the introduction of the single currency.
On the screens of the new European Digital Innovation Center, statistics scroll proudly: 65% of European data is now hosted on our sovereign infrastructures. The digital trade deficit, which reached 80 billion euros in 2023, has transformed into a surplus. Even more impressive, more than 120,000 expatriate engineers have returned to exercise their talents on European soil over the past three years.
In hospitals connected to the European health network, doctors instantly access diagnostic AI models trained on millions of anonymized records, strictly respecting patient consent. In Stockholm, Northern Europe’s largest data center now provides enough recovered heat to warm 30% of the city’s homes, while its carbon footprint has decreased by 40% thanks to “Green Coding” techniques developed in Grenoble and Munich.
“What seemed utopian ten years ago is now our daily reality,” declares the President of the European Commission during her speech in Berlin. “By regaining control of our digital destiny, we have not only created a prosperous economic model, but also a societal model where technology truly serves the citizen.”
From Lisbon to Helsinki, from Dublin to Athens, European digital centers of excellence now form an interconnected network where ideas and innovations flow freely. Universities collaborate in real time on common infrastructures, while European startups now raise funds comparable to their American counterparts.
The transition was not without pain. It was necessary to overcome deep institutional blockages, reform administrations, and sometimes accept short-term costs for long-term benefits. But today, as Europe exports its sovereign and ethical cloud model to other regions of the world, the consensus is unanimous: the game was well worth the candle.
The cold shower of a digital colonization X-Ray
This prosperous vision of digital Europe in 2030 remains for now a mere projection, a possible future but by no means guaranteed. The current reality is quite different, and the numbers speak for themselves.
Imagine if 72% of the French defense budget were paid directly to the American army, or if 72% of our agriculture depended on imports from a single country. This would be considered a national emergency, a sovereignty crisis deserving immediate mobilization. Yet this is exactly what is happening with our cloud.
72% of the French cloud budget flies directly to American giants. This dependency breaks down precisely: 45% for AWS, 18% for Microsoft Azure and 8% for Google Cloud. This is not just dominance; it’s a structured near-monopoly.
This vulnerability takes on a particularly worrying dimension as the Trump administration prepares radical changes in American policy. The “America First” doctrine applied to digital could quickly transform our technological dependence into a real strategic trap. Threats of tariffs, restrictions on access to critical technologies, or worse, instrumentalization of European data, are no longer theoretical hypotheses but concrete risks. The urgency to act has never been so palpable.
This staggering figure is unfortunately only the tip of a much deeper problem. France, which prides itself on having invented the Minitel and having been a pioneer of the Internet in Europe, seems to have abandoned all ambition in this strategic field of cloud computing. How can a country that trains some of the best engineers and mathematicians in the world fail so systematically to create its own digital champions?
This burning question hides a disturbing reality: our technological dependence is not a technical fatality, but the result of questionable political and strategic choices. When we entrust our most sensitive data to companies subject to the American CLOUD Act, it’s not for lack of alternatives, but for lack of long-term vision.
Our digital sovereignty erodes day by day, while our talents expatriate and our technological trade deficit inexorably widens. The time is no longer for observations, but for decisive action. For behind the figures and analyses lies the fundamental issue: do we want to remain masters of our digital destiny, or do we accept becoming a mere digital colony of the American empire?
The vision of a sovereign European cloud that I have outlined is not utopian, however. It is based on technologies that already exist, skills that we master, and political will that only needs to assert itself. The gap between our current situation and this desirable future is not technical – it is strategic and decisional.
In the following pages, I propose a journey to the heart of our digital dependence, its root causes, and especially concrete solutions to escape it. The stake is nothing less than our collective freedom in a world where digital has become the new terrain for power relations between nations.
KEY POINTS
Our digital dependence is not a simple economic statistic, but a major strategic vulnerability: 72% of our cloud budget goes to the United States, our digital deficit is widening, and our talents are massively expatriating. This situation is not a technical fatality, but the result of questionable political choices. The stake is nothing less than our collective freedom in a world where digital has become the terrain for power relations between nations.
France, a digital colony?
France has become a veritable digital colony. We import technology as others once imported cotton, and we export our brains as others exported their raw materials. Our dependence is such that today, an interruption of American cloud services would instantly paralyze France. Public services, CAC 40 companies, administrations… everything now relies on these foreign infrastructures. A major strategic vulnerability at a time when data control conditions the sovereignty of nations.
This situation is all the more frustrating as it is relatively recent. Contrary to the received idea of a French industrial decline that began in the 1970s, our trade balance only truly tipped into the red in 2005. For nearly 35 years, between 1970 and 2004, France maintained a generally balanced position, even showing a surplus during certain periods. The turn of the 2000s coincided with the explosion of digital technology and the rise of GAFAM.
This dependence accelerated with the rise of cloud computing, going from a digital deficit of 35 billion euros in 2010 to more than 80 billion in 2023. A gulf that continues to widen as the French cloud market grew from 1.4 billion € in 2020 to 2.5 billion € in 2022, and is expected to reach 27 billion € by 2025. We are therefore not only losing independence, but also the opportunity to capture this average annual growth of 2.8 billion €.
This chronic deficit is largely fueled by our technological dependence: we now import 79% of our IT solutions, including services, compared to only 42% at the beginning of the 2000s. The paradox is striking: this dependence is taking hold precisely as France trains more than 30,000 engineers each year who are recognized worldwide for their excellence in mathematics and computer science. In fact, American technology companies employ more than 15,000 French engineers in key positions. This brain drain is not anecdotal: one in six French engineers expatriates (16%, or 124,000 out of 780,000). The salary gap largely explains this phenomenon, with an average salary of 65,000€ in France compared to 103,000€ in the United States for equivalent positions. In Paris, a developer earns about 51,000€, while his counterpart in San Francisco earns the equivalent of 134,000€. Not surprisingly, 79% of expatriates cite more advantageous salaries as their main motivation for leaving France.
The VMware case: when the hostage realizes they’re a hostage
The VMware affair perfectly illustrates the dangers of this dependence. This virtualization software, a cornerstone of many cloud infrastructures, was acquired by Broadcom in November 2023 for 69 billion dollars. The consequences were immediate: elimination of perpetual licenses in favor of annual subscriptions, switching from per-processor to per-core billing, and above all, dizzying price increases, up to 300% for some customers, and even 1050% in the extreme case of AT&T.
When Broadcom drastically increased VMware prices, it wasn’t just CIOs who suffered. It was French taxpayer money that had to fill this gap, while essential public services like our hospitals and schools operate on tight budgets.
The result? The entire CAC 40, the State, and French hosts found themselves held hostage. OVH’s private cloud and even Scaleway used this technology. Many forgot that when you put all your eggs in one basket, you need to know who’s holding the basket. This strategy, described as “wringing out” by some analysts, shows the risks of a technological monopoly: Broadcom, initially a chip manufacturer, can now impose its conditions on the entire digital ecosystem, forcing many organizations to consider alternatives such as HyperV, OpenNebula or Proxmox.
This dependence also exposes France to considerable security risks. The figures are alarming: 47% of business data hosted on the cloud is considered sensitive, while 44% of organizations have already experienced a cloud data breach. More worryingly, cloud vulnerabilities doubled between 2019 and 2023, from 1,700 to 3,900. In total, 79% of French companies have experienced at least one cloud security breach since 2020, a statistic that should give our public and private decision-makers pause for thought.
When pipes are confused with content
Confusing the cloud with infrastructure is like confusing a television channel with its antenna. Orange is excellent at installing antennas, but it’s not Orange that creates the content that passes through them. The cloud is primarily a set of intelligent software that runs on infrastructure, not the other way around.
The cloud is systematically approached as an infrastructure question, when it is fundamentally a software layer. Let’s take a moment to deconstruct this misconception: if the cloud were just a matter of infrastructure, the giants of the sector would be called AOL and Verizon in the United States, or Orange in France. Yet it is software companies that dominate this market, not infrastructure ones.
This confusion has led to disastrous strategic choices. The Andromède project (Cloudwatt, Numergy), launched in 2012 with 150 million euros of public investment, perfectly illustrates this error. The State, rather than supporting innovative players already in the market like OVH, Clever Cloud or Scaleway, preferred to inject hundreds of millions into established behemoths: Orange, SFR, Bull, Thales.
The ambitions were excessive: 500 million euros in revenue by 2017 for Cloudwatt, 400 million by 2016 for Numergy. The reality was quite different, barely 2 million for Cloudwatt and 6 million for Numergy in 2014. As Jules-Henri Gavetti, CEO of Ikoula, summed it up: “It’s wasted time. These companies generated buzz without really educating the market. Meanwhile, American players have become more powerful.” The project ultimately ended with Orange taking over Cloudwatt and SFR taking over Numergy, quietly burying the ambition of a sovereign French cloud.
These projects never came to fruition, but they siphoned the best talent from innovative startups to put them at the service of directions without a clear technological vision. These resounding failures have durably tarnished the image of the French cloud, creating the persistent impression that our companies “don’t know how to do it” – even though they weren’t at the helm.
French gems sacrificed on the altar of bureaucracy
Contrary to popular belief, France does innovate. OVH has developed its own server cooling system that is more efficient than Amazon’s. Clever Cloud created container orchestration technology even before Docker popularized the concept. Even in cutting-edge technologies like distributed databases, gems like Scality compete with American giants.
France therefore does not lack innovative companies in the cloud domain. OVH (created in 1999), Scaleway (formerly Online), or Clever Cloud offer technically advanced solutions. But these players remain marginalized compared to American giants for several reasons.
The war of French egos versus the strategic diplomacy of GAFAM
First, a chronic inability to collaborate, which sometimes borders on collective self-sabotage. Where GAFAM knows how to set aside their differences to agree on strategic issues (such as the contract between Google and Apple to be the default search engine on iPhone, which represents a third of Apple’s profits, or about 20 billion dollars annually), our champions exhaust themselves in internal quarrels and non-constructive competition.
This fragmentation can be observed at several levels. At the technological level, each French player systematically reinvents the wheel rather than building on the successes of others. When Scaleway develops an innovation in server cooling, OVH prefers to design its own solution rather than considering a cross-license or partnership. When Clever Cloud offers to deploy its PaaS platform on the infrastructure of other French hosts to enrich their service catalogs, the answer is invariably ‘No, we’ll code it ourselves’, a promise rarely kept that ultimately leaves the field open to Americans.
At the commercial level, this inability to collaborate manifests in the absence of joint offerings for large accounts. While AWS and Microsoft offer integrated ‘all-in-one’ solutions to large French groups, our champions remain unable to present a united front, forcing CIOs to multiply suppliers if they want to remain in the French ecosystem, an administrative complexity that few are willing to take on.
The causes of this situation are multiple. On one hand, a French entrepreneurial culture that values technical innovation more than building strategic alliances. On the other hand, a financing ecosystem that puts our champions in competition for limited resources rather than encouraging synergies. Finally, the absence of an industrial vision carried by the State, which could play a facilitating role between these actors.
The contrast with the American ecosystem is striking. In Seattle, headquarters of the three main American clouds, talent circulates freely between Amazon, Microsoft and Google, creating a mixing of ideas and practices that feeds collective innovation. These giants, while engaging in fierce competition, actively collaborate on strategic open-source projects like Kubernetes, and regularly ally against regulators to defend their common interests. They understand that their strength lies as much in their individual excellence as in the power of their collective ecosystem.
Gaia-X: a promise diluted by American hyperscalers
Faced with these repeated failures, the Gaia-X initiative finally seemed to propose a radically different and promising approach. Rather than trying to build infrastructure competing with American hyperscalers, Gaia-X focuses on creating a framework for interoperability and open standards. This software approach allows guaranteeing the portability of data and applications between different cloud providers, ensuring interoperability between different platforms, and defining data sovereignty standards allowing European companies to maintain control over their sensitive information.
With more than 300 members, including major European users and all important cloud service providers, Gaia-X finally seemed on the right track. Its governance was even built to meet European requirements, limiting the board of directors to entities with headquarters in Europe.
But why isn’t this initiative more prominently featured at the European level? The answer is as simple as it is desperate: the project has been literally “diluted” by the entry of American giants into the association. Frank Karlitschek, CEO of NextCloud and founding member of Gaia-X, even stopped attending meetings, believing that the project had been “sabotaged from within” by the hyperscalers themselves.
Added to this infiltration is excessive bureaucracy typical of European initiatives. While the American DARPA operates with almost total decision-making autonomy granted to teams, Gaia-X has become entangled in multi-layered administrative structures that paralyze any quick decision-making.
More worryingly, the very vision of the project has been transformed. Initially conceived as a European “virtual hyperscaler,” Gaia-X has morphed into a simple certification and standards system, thus losing part of its initial ambition. As Max Claps, research director at IDC, explains, the project “quickly transformed into an international initiative less focused on protecting the sovereignty of European digital infrastructure, and more on creating models that become the global reference for digital sovereignty.” A nice way of saying that the wolf has entered the sheepfold and rewritten the rules to its advantage.
For Gaia-X to truly become the spearhead of the European sovereign cloud, it would be necessary to reaffirm the independence of its governance by strictly limiting the influence of non-European actors, simplify its structure by drawing inspiration from the DARPA model, and obtain a strong political mandate at the highest European level. But do we still have the political will to defend our digital sovereignty, or do we prefer to continue to delude ourselves by signing checks to American giants?
When the French State finances its own chains of dependence
Next, a totally dysfunctional public procurement that acts as a systemic lock against French solutions. The problem goes far beyond simple preferences: specifications are often written in such a way as to exclude national actors from the outset, through technical specifications that exactly match American offerings.
The Health Data Hub (HDH) scandal is a perfect illustration and deserves closer attention. In 2019, the French State decided to centralize all French health data, an informational asset of inestimable strategic value. Without a public call for tenders, the project was directly awarded to Microsoft Azure. The justifications advanced by Stéphanie Combes, director of the HDH, were that only Microsoft could offer the necessary scale and required security guarantees. These arguments were strongly contested by French actors, who did not even have the opportunity to demonstrate their capabilities. Even more troubling: those responsible for the project never explained why a hybrid solution, combining several French providers to achieve the required scale, was not considered.
When the Council of State was finally seized on this matter, it recognized the risks of infringement on personal data protection, but only ordered a vague “compliance” without questioning the choice of provider. This decision revealed a profound lack of understanding of technological issues at the highest French jurisdictional level.
This problem is systemic and affects all strata of the administration. The public procurement code itself constitutes a major obstacle, as it was designed for the purchase of material goods and not innovative technological services. Its selection criteria invariably favor established solutions at the expense of disruptive innovations. For example:
- The requirement for similar references on previous projects, which mechanically favors historical actors
- Financial criteria that often impose a minimum turnover equivalent to several times the value of the market
- The obligation to demonstrate international intervention capacity, an impossible criterion to satisfy for French startups in growth phase
- Excessively long decision cycles (12 to 18 months) that require cash flow that only large groups can support
The DINUM (Interministerial Digital Direction), supposed to guide the State’s digital strategy, paradoxically contributes to the problem by developing its own cloud solutions rather than supporting the French private ecosystem. This unfair competition from the State against its own companies creates additional confusion and further fragments national efforts.
The consequences are disastrous on several levels. First, a loss of sovereignty over critical data, as shown by the invalidation of the Privacy Shield in July 2020 by the Court of Justice of the European Union, which made technically illegal the hosting of French data on American clouds subject to the CLOUD Act. Next, a vicious circle where French actors, deprived of structuring public markets, struggle to reach the critical size that would precisely allow them to meet the requirements of tenders.
Alternatives exist, however. The American Small Business Act reserves a share of public contracts for innovative SMEs. The United Kingdom has put in place a system of “open by default public markets” where technical specifications are developed in consultation with the local ecosystem. Germany, for its part, has developed a modular approach allowing several suppliers to be combined on the same project, thus facilitating the integration of specialized actors of modest size.
The necessary reform is deep and urgent. It would involve not only an overhaul of the public procurement code, but also training public buyers on the strategic issues of digital technology, and implementing concrete incentives to favor sovereign solutions.
Faced with these obstacles, French clouds must also confront intense and permanent lobbying from American giants, whose influence can easily be measured at sector conferences.
The financial absurdity of our under-investment
The financial absurdity of our under-investment is all the more glaring when measuring the gap with our competitors. The investment gap between the EU and the United States in ICT and cloud reaches 1.36 trillion dollars. Even more revealing, American startups raised 190 billion dollars in 2024 (+25%), compared to only 7.8 billion euros for French startups (-7%). In AI specifically, American startups raised 38 billion dollars in 2024 compared to only 4.1 billion for all European startups combined. Faced with such gaps, how can we be surprised by our lag?
The DARPA Model Without Innovation, But With Bureaucracy
In the face of these repeated failures, what model could inspire a French digital renaissance? The answer has existed for more than 60 years: the American DARPA. Created in 1958 by the Department of Defense, this agency has revolutionized our world by funding technologies that gave birth to the Internet, GPS, and even autonomous vehicles. But it’s not so much its successes as its method that should inspire us.
The philosophy of DARPA rests on a radical principle: planned failure. The agency deliberately funds “moonshot” projects with an assumed failure rate of 80%. Even more revealing: when this rate decreases, leaders worry – not about wasting public money, but about not taking enough risks! This culture of boldness relies on three pillars: entrusting projects to technical experts and not managers, limiting their duration (generally 4 years) to maintain urgency, and granting almost total decision-making autonomy to teams.
Europe tried to draw inspiration from this model by creating JEDI (Joint European Disruptive Initiative). On paper, the ambition was laudable: stimulate breakthrough innovation in strategic sectors such as climate, health, and digital to compete with the United States and China. But the copy transformed into a caricature.
Where DARPA embraces risk and autonomy, JEDI became entangled in the typically European bureaucratic straitjacket: interminable selection processes, multi-layered governance making any rapid decision impossible, and budgets scattered in ineffective sprinkling. Instead of being an innovation accelerator, JEDI became yet another example of our inability to translate good ideas into concrete actions – one more administrative machine that produces more reports than disruptive technologies.
We studied the DARPA model, we admired it, then we did exactly the opposite of what made it successful. Our European version perfectly illustrates what physicist Richard Feynman called “cargo cult science”: we reproduced the external forms without understanding the deep mechanisms that generate results.
Spectator of Our Own Missed Revolution
The Villani report, published in 2018, should have been our masterstroke in the global race for artificial intelligence. This visionary document proposed an ambitious and coherent national strategy: massive investments in research, overhaul of our education system, and concentration of efforts on four strategic sectors: health, environment, transport, and defense. With just 200 million euros per year, a drop in the bucket of the State budget, we could have laid the groundwork for a first-class French AI ecosystem.
But what did we do with this report? We celebrated it, applauded, then carefully placed it on a shelf where it continues to gather dust. Meanwhile, the United States invested massively, China deployed its national strategy, and even countries like Israel or South Korea took bold initiatives. We remained on the platform, watching the innovation train pass without ever boarding.
The irony is cruel: while we hesitated to invest 200 million euros annually, OpenAI raised billions of dollars and revolutionized the sector with ChatGPT. French talents participated in this revolution, but under the American flag. Today, we marvel at these advances as if they were the fruit of inaccessible magic, even though we had in our hands the recipe to create our own “FrenchAI” two years before the emergence of OpenAI.
This paralysis in the face of action, this chronic inability to transform reports into concrete achievements, perfectly illustrates our technological malaise. We excel in diagnosis, the analyses of the Villani report were brilliant, but we seem allergic to treatment. As if we preferred to remain indefinitely in the waiting room rather than face the operation that could cure us.
Our recent history in AI is one of a missed appointment with History, a golden opportunity transformed into eternal regrets, a potential wasted by lack of political audacity and industrial vision. While others were building the future, we continued to debate its relevance.
The “Not Invented Here” Syndrome in Reverse
We suffer from a strange syndrome: while Silicon Valley sometimes rejects external solutions out of pride, France systematically rejects its own innovations due to an inferiority complex. An absurd paradox when observing the massive presence of French people in major global AI projects. Our graduates from École Normale Supérieure and École Polytechnique are everywhere, at Google, Meta, OpenAI, except at the helm of our own strategic projects. This pathological distrust of our own technology is not natural, but the fruit of perfectly orchestrated American soft power since the post-war period.
This French malaise finds its roots far in our history. Few know it, but in exchange for Marshall Plan aid (American economic aid program launched in 1947 to rebuild Europe after World War II, injecting about 13 billion dollars over four years (1948-1952) to stimulate economic recovery and counter Soviet influence), European countries had to accept economic and cultural conditions, including opening their markets to American films. A soft power strategy of formidable effectiveness that has progressively instilled the idea that technology could only be American.
This technological inferiority complex is, however, flagrantly absurd when observing the massive presence of French people in major global artificial intelligence projects. Graduates from École Normale and École Polytechnique, our brilliant minds are everywhere, optimizing Google’s algorithms, directing research at DeepMind, designing OpenAI’s architecture, everywhere except at the helm of our own strategic projects.
France presents a striking paradox: we train elite engineers who revolutionize global technology, but entrust the governance of our national digital projects to énarques and senior officials trained in administration, not innovation. Where Silicon Valley places PhDs in artificial intelligence at the head of transformative projects, we appoint Sciences Po and ENA graduates to lead our digital strategy.
This institutionalized cognitive dissonance largely explains our repeated failures. A technological project led by someone who has never written a line of code resembles an orchestra conducted by someone who has never played an instrument: technically possible, but rarely harmonious. Our administrative elites, brilliant in their field, find themselves piloting transformations of which they master neither the foundations nor the language, with predictably catastrophic results.
Even more revealing: when a French Polytechnique graduate creates an innovative startup, he is often imposed a “business CEO” by his investors, often from the same top business and administration schools. France systematically prefers administrative governance to technical expertise, even when recent history demonstrates the failure of this approach. We continue to apply the recipes of a bygone industrial era to the knowledge economy. A pattern that repeats itself with disconcerting regularity.
Three major obstacles paralyze our digital sovereignty: 1) A persistent confusion between infrastructure and software, which leads to poorly targeted investments; 2) A chronic inability of French actors to collaborate effectively, unlike American giants who know how to ally on strategic subjects; 3) A dysfunctional public procurement that systematically favors foreign solutions at the expense of French innovations. These blockages are not technical, but organizational and cultural.
An opaque billing model that traps companies
The myth of the cheaper American cloud is particularly persistent. In reality, the price difference between a millisecond of computation at OVH and at Amazon can reach a factor of 8 or 9. Precise comparisons confirm this: OVHcloud offers storage rates twice as cheap (0.002€/GB compared to $0.004/GB for AWS), while Scaleway offers small virtual instances at $10.83/month compared to $20.49/month at Google Cloud. Paradoxically, public cloud services in Europe are on average 7-19% more expensive than in the United States, an extra cost that is explained notably by the overcharging practiced by American giants in the European market.
The price difference between French and American cloud services is explained by a radically opposing billing philosophy. Imagine two restaurants: the first charges you for the size of the table you reserve (French approach), the second bills you for each bite consumed, each minute spent, and adds fees if you talk too loudly (American approach).
French hosts have adopted a simple and predictable pricing model: they mainly bill for reserved capacity, what could be called “the size of the pipe.” You pay for certain defined computing power and storage space, whether you use them completely or not. This approach allows companies to precisely budget their cloud expenses for the year, without unpleasant surprises.
Conversely, American giants like AWS have developed a much more complex model that bills for every aspect of usage: the volume of data that circulates, the number of API calls, the consumed computing time, input/output operations, and even sometimes technical support. While this approach may seem fairer in theory (“you only pay for what you consume”), in practice it makes any budget forecasting virtually impossible.
This fundamental difference creates a disconcerting paradox: in my interviews with IT directors, they frequently cite “savings” as the main reason for choosing AWS or Azure. Yet these same managers often admit, after a few drinks, having experienced budget overruns of 30 to 50% compared to initial estimates. The final bill almost always turns out to be significantly higher than expected, and well above the proposals of French actors.
French hosts persist in their transparent billing model out of conviction as much as strategy: they consider that an accessible, predictable web without surprises promotes innovation and democratization of cloud technologies. This vision, although virtuous, struggles to counter the sophisticated marketing of American giants who boast artificially low entry prices, knowing full well that costs will explode with actual usage.
It’s as if we had collectively succumbed to an economic mirage, attracted by the promise of fresh water in the desert, only to discover too late that it was a salt sea billed by the sip.
A software approach to the sovereign cloud would allow developing cost management and optimization tools specifically adapted to the European context, creating standardized interfaces making it easy to compare offers from different providers, and setting up automatic expense control and limitation mechanisms. These tools would give European companies the means to make informed choices, based on the real and total cost of ownership, rather than on misleading marketing promises.
The software approach, key to recovered sovereignty
If we truly want to regain our digital sovereignty, we must stop confusing the cloud with infrastructure and understand that it is primarily a software issue. This approach error has led to repeated failures and growing dependence on American giants.
Software sovereignty and standardized interoperability
The concept of software sovereignty constitutes one of the three essential pillars of the sovereign cloud: “Cloud services must be usable, regardless of the provider. Thus, migration to another infrastructure must be facilitated and above all possible.”
This approach allows breaking the lock-in mechanisms put in place by American giants, which make changing providers extremely costly and complex. By focusing on the software layer and imposing interoperability standards, Europe can regain its technological independence without necessarily owning the underlying infrastructure.
Interoperability is the keystone of a European sovereign cloud ecosystem. To guarantee it, Europe must impose strict standards, drawing inspiration from the REST API (REpresentational State Transfer) model that revolutionized web development.
A major advance in this direction is the recent creation of SECA (Sovereign European Cloud API), an initiative launched in March 2025 by European providers Aruba and IONOS, in collaboration with the Dynamo cloud marketplace. This European sovereign API specifically aims to allow different European cloud providers to offer their customers the ability to run applications and workloads on different clouds, to eliminate the vendor lock-in problem, and to remain compliant with European rules on data sovereignty, AI, and data protection.
As Achim Weiss, CEO of IONOS, explains: “AI and Cloud are transforming the global economy, and Europe cannot afford to be left behind. Europe needs a strong and sovereign digital ecosystem. SECA is a crucial step in building a secure, independent, and sustainable digital infrastructure – an infrastructure that keeps Europe strong, competitive, and in control.”
A concrete European roadmap, not a mere declaration of intent
The realization of this software sovereignty requires precise and measurable actions, not new commissions or vague reports. We must break the infernal cycle of findings without action.
- First step: impose open source as the foundation of the sovereign cloud. Not out of ideology, but out of strategic pragmatism. Europe must invest massively – at least 2 billion euros over 5 years – in critical software components such as virtualization systems, container orchestration, and distributed storage. These technologies form the backbone of any modern cloud. Projects like OpenStack, European Kubernetes, or Ceph show that European technical excellence already exists, but suffers from chronic underfunding compared to the billions injected by GAFAM into their proprietary alternatives.
- Second step: transform regulatory constraint into competitive advantage. The GDPR has shown the way. It’s time to go further with a European “Cloud Compatibility Act” that would impose total interoperability between clouds and prohibit technical lock-in practices. Concretely, an application developed for AWS should be able to work without major modification on OVH or Scaleway. This constraint would force American hyperscalers to open their technologies or progressively lose their grip on the European market.
- Third step: create a European institute for sovereign cloud software, directed not by bureaucrats but by elite engineers recruited under international market conditions. This institute would combine a DARPA model for funding innovative projects and a technical certification laboratory that would guarantee compliance with interoperability standards. Its annual budget – at least 500 million euros – would be a drop in the ocean compared to the cost of our current dependence, but a strategic investment for our future independence.
- Fourth step: develop a suite of automated migration tools to facilitate the transition of existing systems to sovereign solutions. These tools would automatically analyze the dependencies of current cloud applications and propose migration paths to open alternatives. The experience of companies like Clever Cloud demonstrates that this approach is technically feasible, but requires a coordinated industrial effort that only a European initiative can carry.
- Fifth step: create a European mechanism for protection against predatory acquisitions. The VMware case could have been avoided if a European rapid intervention fund had existed to counter Broadcom’s offer. For each strategic cloud technology developed in Europe, we must be able to mobilize within a few weeks the necessary capital to prevent its passage under foreign control. This hybrid public-private mechanism could draw inspiration from the Strategic Investment Fund model, but with increased reactivity and a mandate specifically centered on cloud technologies.
This approach would imply a radical change in our way of approaching strategic digital projects. Gone are the grand five-year plans without concrete results. Instead, an agile method, where every euro invested must produce measurable results within 12 months. Each project would be evaluated not on its intentions, but on its effective contribution to reducing our technological dependence, with quantified objectives for decreasing the penetration rate of American hyperscalers in the European market.
The VMware affair has brutally awakened us. It could be the last wake-up call before collapse, or the catalyst for a European digital renaissance. The difference lies in our ability to transform this awareness into concrete and immediate actions.
European digital sovereignty passes through a fundamentally software approach, articulated around five axes: 1) The massive adoption of open source as a technical foundation; 2) The transformation of regulatory constraints into competitive advantages; 3) The creation of a European cloud software institute led by technical experts; 4) The development of automated migration tools; 5) A European mechanism for protection against predatory acquisitions. This approach requires less financial investment than an infrastructure strategy, but demands a radical break with our current methods.
Energy optimization through software
Energy efficiency represents a major competitive advantage for Europe in the cloud domain. Companies like OVH have developed server cooling systems more efficient than those of Amazon. But beyond infrastructure, it’s especially at the software level that the greatest energy savings can be achieved.
Concrete examples of energy optimization through software
The partnership announced in April 2025 between Energisme, an energy intelligence platform, and NumSpot, a player in the French sovereign cloud, perfectly illustrates how the software approach can generate significant energy savings. This collaboration aims to offer energy consumption optimization services in a sovereign cloud environment qualified SecNumCloud. As Thierry Chambon, Managing Director of Energisme, explains: “This partnership with NumSpot is a decisive step in our quest for an intelligent and secure energy transition. Together, we’re creating an environment where companies can fully exploit the potential of their energy data in a sovereign framework, hosting and computing, totally protected.”
This partnership enables real-time analysis of cloud infrastructure energy consumption, automatic optimization of resources according to real needs, and prediction of consumption peaks to anticipate and distribute the load.
European companies like Qarnot Computing have developed sophisticated algorithms that allow distributing computing loads according to the available energy mix. Concretely, these software solutions can automatically move non-urgent workloads to periods when electricity is most decarbonized, geographically distribute calculations according to available energy sources (for example, favor Nordic data centers powered by hydroelectricity during periods of low solar production), and dynamically adapt the allocated computing power according to the urgency of tasks and energy cost.
European containerization technologies, like those developed by Clever Cloud, allow a much more efficient use of hardware resources, with increased density of workloads on the same physical server, automatic starting and stopping of instances according to needs, and dynamic allocation of resources to avoid oversizing.
European startups like Submer or Asperitas have developed software solutions that optimize immersion cooling systems in real time, with continuous adjustment of cooling parameters according to the load, prediction of cooling needs based on workload analysis, and intelligent heat recovery for other uses such as district heating or agricultural greenhouses.
Green Coding
Beyond infrastructure optimization, an approach centered on “Green Coding” would allow considerably reducing the energy footprint of European cloud applications. This approach consists of fundamentally rethinking development methods and algorithms to minimize energy consumption.
Traditional algorithms have often been designed by prioritizing execution speed without consideration for energy efficiency. Research conducted by European universities such as the École Polytechnique Fédérale de Lausanne has demonstrated that redesigning fundamental algorithms can reduce energy consumption by 30 to 70%. This involves replacing classic sorting algorithms (quicksort, mergesort) with variants optimized for energy consumption, using compression algorithms adapted to the nature of data rather than generic solutions, and implementing network routing algorithms that minimize hops and optimize paths according to the energy efficiency of equipment.
Specific development practices can also significantly reduce energy consumption: reduction of memory read/write operations, which are particularly energy-intensive, optimization of data structures to minimize memory footprint, use of intelligent caching techniques that reduce redundant calculations, and adoption of functional programming that favors immutability and reduces resource-costly side effects.
Europe could develop and promote frameworks specifically designed to minimize energy footprint, such as AI libraries optimized to reduce the energy consumption of models (like EcoLLM, a French initiative that reduces the carbon footprint of large language models by 30 to 50%), web frameworks that minimize data transfers and optimize client-side rendering, and development tools that integrate energy efficiency metrics into the development cycle.
Specific optimization tools could automatically transform existing code to make it more energy efficient, with intelligent compilers that generate machine code optimized for energy consumption rather than pure speed, static analysis tools that identify energy-intensive portions of code, and continuous optimization systems that adjust code according to actual usage patterns.
A concrete example is the Green Algorithms project from the University of Cambridge, which demonstrated that a simple optimization of medical image processing algorithms allowed reducing energy consumption by 83% while maintaining the same quality of results. Applied at the scale of the European cloud, this approach could generate energy savings of several terawatt hours per year.
To go further in energy optimization through software, Europe could create a European standard for measuring cloud energy efficiency, develop a European platform for sharing energy data, integrate electrical grid constraints, and create an “energy passport” for cloud applications.
According to available data, adopting an optimized cloud approach can reduce the carbon footprint of IT services by 30 to 90% compared to traditional infrastructures. By developing software solutions specifically designed to optimize resource use, Europe could not only reduce its energy costs but also position itself as a world leader in eco-responsible cloud computing.
Resource pooling for considerable economies of scale
One of the major advantages of the cloud is resource pooling. However, this pooling is only effective if it is orchestrated by an intelligent software layer. A software approach to the sovereign cloud would allow for considerable savings at several levels.
Concrete examples of achievable savings
According to a study by Cigref (Club Informatique des Grandes Entreprises Françaises), pooling storage and computing resources in a sovereign cloud could generate economies of scale of 30 to 45% compared to isolated infrastructures. Concretely, a data center shared among several European public administrations could reduce infrastructure costs by 42% compared to separate data centers. Server consolidation via virtualization allows reaching utilization rates of 60-70% versus 15-20% for dedicated servers, multiplying investment efficiency by 3 to 4 times. Sharing maintenance and security costs among several entities reduces operational expenses by 35% on average.
Pooling also allows negotiating more advantageous license contracts. Group purchasing of licenses for European public administrations has allowed documented savings of 25 to 40% in pilot projects. The development of pooled open source solutions completely eliminates licensing costs, with potential savings of 60 to 80 million euros per year for French administrations alone. Usage-based billing models optimize costs according to actual needs, with average savings of 22% compared to perpetual licenses.
Pooling also optimizes resource utilization and reduces energy consumption. Modern pooled data centers achieve a PUE (Power Usage Effectiveness) of 1.1 to 1.2, compared to 2.0 to 2.5 for isolated infrastructures, representing a 50 to 60% reduction in energy consumption. Optimizing workload between different European data centers according to local climate conditions and energy mix can reduce carbon footprint by 40%. Large-scale heat recovery becomes economically viable, with projects like Stockholm Data Parks reusing data center heat to warm 10% of the city’s homes.
Pooling also allows sharing rare and costly skills. A security team shared among several organizations can reduce costs by 30 to 50% while improving the level of protection. Pooled operations teams can manage 3 to 4 times more servers than a team dedicated to a single organization. Sharing specialized skills (AI, big data, DevOps) allows small structures to access expertise otherwise inaccessible.
Finally, pooling regulatory compliance efforts generates significant savings. The development of pre-certified GDPR or SecNumCloud solutions reduces compliance costs by 60 to 70% for each organization. Shared security audits reduce costs by 40 to 50% per organization. Shared regulatory monitoring allows anticipating changes and reducing adaptation costs.
A concrete example is the MarioNUM project, a pooled educational cloud for French universities, which reduced infrastructure costs by 38%, license costs by 45%, and energy consumption by 52% compared to individual solutions previously used by each institution.
To maximize these savings, a software approach to the sovereign cloud would allow developing more efficient resource allocation algorithms, adapted to European usage patterns, creating resource sharing mechanisms between different administrations and public companies, and setting up intelligent workload prioritization systems according to their strategic importance.
What if we took the example of a “Cloud Digital Euro”
Just as the single currency transformed the European economy, could we imagine a “Cloud Digital Euro” that would unify the European approach and create a credible alternative to American hegemony? This analogy is particularly relevant and could constitute a major breaking point.
The parallels between the monetary euro and a potential “Cloud Digital Euro”
The Euro was created to strengthen European economic integration and offer an alternative to the American dollar. Similarly, a “Cloud Digital Euro” could aim to unify the European digital market, offer an alternative to dominant American solutions, strengthen European sovereignty, and create a common governance framework.
Just as the Euro eliminated frictions related to exchanges between national currencies, a European cloud standard would eliminate technical barriers between different national cloud providers. This unification would allow the creation of a true single cloud market with more than 450 million potential users, the elimination of transaction costs related to switching from one provider to another, and the standardization of interfaces and protocols, facilitating interoperability.
The Euro is today the second global reserve currency after the dollar. A standardized European cloud could become a credible alternative to American hyperscalers. This alternative would respond to the specific needs of European companies and administrations, offer guarantees of sovereignty and data protection impossible to obtain with American providers, and create an innovation ecosystem adapted to European values and priorities.
The Euro gave Europe greater monetary autonomy. A sovereign cloud would give Europe greater digital autonomy. This sovereignty would translate into independence from extraterritorial laws like the Cloud Act, protection against economic sanctions that could target access to digital services, and control over critical infrastructures and strategic data.
The European Central Bank oversees the Euro. A “European Cloud Authority” could oversee the standards and compliance of the European cloud. This governance would establish common rules for all market players, ensure fairness and transparency in access to resources, and ensure the stability and resilience of the European cloud ecosystem.
Just as the Euro embodies certain European values in its very design, a European cloud could embody the values of privacy protection, transparency, and fairness that are at the heart of the European project. This model would place personal data protection at the center of its design, promote algorithmic transparency and system explainability, and integrate ethical and environmental considerations from the design stage.
The Euro has helped strengthen Europe’s economic position on the world stage. A European cloud could stimulate technological innovation and strengthen the competitiveness of European companies. This stimulation would create fertile ground for the emergence of European technological champions, promote the development of solutions adapted to the specificities of European industries, and allow Europe to retain the added value generated by the digital economy.
The analogy with the Euro is all the more relevant as, as Max Kretschmer, press attaché at Finance Watch, explains: “The digital euro is no longer just a potential new payment method. It is becoming increasingly crucial for Europe’s strategic autonomy.” In the same way, a “Cloud Digital Euro” would not simply be a technical alternative, but a crucial element of European strategic autonomy in the digital domain.
The conditions for success of a “Cloud Digital Euro”
To succeed, this initiative would require adapted governance, inspired by both the ECB for the Euro and web standardization bodies like the W3C.
At the technical level, a “European Cloud Standards Board” would bring together recognized experts, independent of immediate commercial interests, responsible for defining and evolving standards. This council would not be composed of national representatives or theoretical experts, but engineers who have proven their excellence in creating and operating cloud technologies. Its legitimacy would come from its indisputable technical expertise, not its political weight.
At the strategic level, a “European Cloud Authority” (ECA), endowed with real decision-making power and a substantial budget (minimum 1 billion euros annually), would pilot the deployment of the strategy. This authority would have three missions:
- Fund strategic innovation projects in the sovereign cloud domain
- Coordinate the progressive migration of administrations to certified solutions
- Ensure compliance with standards and certifications by all actors
At the political level, a “European Digital Sovereignty Council” bringing together heads of state and government would define strategic orientations and arbitrate sensitive issues related to technological independence. This council would meet at least twice a year, testifying to the importance given to this subject at the highest level.
Support mechanisms and incentives for adoption
To accelerate the adoption of this standard, several levers could be activated:
- A digital transition fund of 10 billion euros over 5 years would finance migration projects of administrations and companies to sovereign solutions. This fund would cover up to 70% of migration costs for SMEs and 50% for large companies, considerably reducing the financial barrier to entry.
- A massive training program would aim to certify 100,000 experts in sovereign cloud technologies by 2030. This program would associate universities, engineering schools, and professional training centers to create a pool of skills adapted to market needs.
- Targeted tax advantages would encourage early adoption of sovereign solutions, with a “Digital Sovereignty” tax credit covering up to 30% of expenses related to migration to certified solutions.
- A digital border adjustment mechanism would progressively apply additional taxation to cloud services not respecting European standards, on the model of what is envisaged for products not respecting our environmental standards.
The analogy with the Euro is all the more relevant as, as Max Kretschmer, press attaché at Finance Watch, explains: “The digital euro is no longer just a potential new payment method. It is becoming increasingly crucial for Europe’s strategic autonomy.” In the same way, a “Cloud Digital Euro” would not simply be a technical alternative, but a crucial element of European strategic autonomy in the digital domain.
An ambitious but realistic timeline
If Europe wants to retain even a chance of regaining its digital sovereignty, it can no longer afford the usual procrastination. Endless consultations, reports without follow-up, and diluted initiatives have already cost our continent a crucial decade. The time is no longer for designing new theoretical roadmaps, but for implementing a clear strategy, marked out with concrete and measurable stages.
A precise calendar, with binding objectives and clearly defined responsibilities, must be established and respected with the same rigor that accompanied the deployment of the monetary Euro:
- 2025: Adoption of SECA standards as an official European norm, with the simultaneous creation of a technical committee dedicated to their evolution. This first fundamental step lays the common framework essential to any future interoperability. European authorities will need to mobilize all actors in the sector to ensure a solid technical consensus.
- 2026: Launch of the digital transition fund (10 billion euros) and deployment of the certification program to train 100,000 experts in sovereign cloud technologies. These two parallel initiatives aim to remove the two main obstacles to adoption: financing and skills. Without this double impetus, even the best standards would remain a dead letter.
- 2027: First “EU Cloud Compatible” certifications awarded to solutions fully respecting interoperability and sovereignty standards. This label, managed by the European Cloud Authority, will progressively become a prerequisite for accessing sensitive public markets. The first certified European champions will benefit from increased visibility and privileged access to European funding.
- 2028: Obligation for all new European public contracts to include a minimum of 50% certified sovereign solutions. This measure, probably the most disruptive of the calendar, will force the entire ecosystem to adapt. For the first time, American giants will have to either truly comply with European requirements or lose ground.
- 2029: Effective interconnection of all certified clouds, allowing users to move their data and applications without friction between different providers. This unprecedented mobility will fundamentally transform the competitive dynamics of the market, drastically reducing the lock-in power of dominant hyperscalers.
- 2030: Official launch of the “Cloud Digital Euro,” marking the culmination of five years of coordinated efforts. This symbolic step will consecrate the emergence of a true European single cloud market, as fluid and integrated as our monetary market has become since the Euro. Users will then be able to navigate between different providers as easily as they use different European banks today.
- 2035: Objective of 75% of public services and 50% of private critical services hosted on sovereign solutions. This long-term ambition will require constant work to improve and adapt European solutions to maintain their competitiveness against foreign alternatives. At this stage, Europe should have a mature, innovative, and resilient cloud ecosystem.
Each year of delay costs us billions in technological dependence and irreparably weakens our strategic position. The monetary Euro took more than 40 years to materialize, from the creation of the European Monetary Snake in 1972 to the introduction of bills and coins in 2002. We do not have the luxury of having such a deadline for our digital sovereignty – in just 10 years, dependence could become practically irreversible, as migration costs increase exponentially with time.
This calendar is not simply an administrative formality, but the backbone of a reconquest strategy. Each milestone represents a potential tipping point in our dependency relationship with non-European providers. Without this temporal discipline, even the best intentions and the most generous funding will disperse over time, losing their transformative effect.
The “Cloud Digital Euro” could unify the European approach as the Euro did for our economy. This vision relies on three-level governance (technical, strategic, and political), powerful incentive mechanisms (transition fund, training, tax advantages), and an ambitious but realistic timeline (2025-2035). Each year of delay costs Europe billions and irreparably weakens our strategic position. This approach would allow Europe to create an alternative model to the American-Chinese duopoly, in line with our data protection values and our concern for energy efficiency.
In clear, do we really want a sovereign cloud?
After this long autopsy of our failures, a fundamental question imposes itself: does France really want a sovereign cloud? Not in words, speeches are never lacking, but in concrete actions, budgetary choices, political courage.
For here is the French paradox: contrary to the defeatist litany we have been hearing for a decade, creating a competitive French sovereign cloud is perfectly achievable on a technical level. Better yet, thanks to the evolution of open source technologies and the maturity of development tools, this objective would require much less investment today than it did ten years ago.
As we have seen, the real question is not that France (or Germany, Italy, Spain, or another European country) sets up its own sovereign cloud, but that all of Europe mobilizes to build a truly strong and sovereign cloud at the continental level.
The stakes go far beyond the economic or technological framework; they are political. In a world where digital now shapes global geopolitics, where mastery of data conditions military power, cultural influence, and economic resilience, digital sovereignty has become an existential question for our continent.
The software approach to the sovereign cloud represents a unique opportunity for Europe to regain its technological independence while achieving significant financial and energy savings. By focusing on creating an open, interoperable software ecosystem that respects European values, rather than building infrastructure to compete with American giants, Europe can transform its apparent weakness into a strategic advantage.
This approach, however, requires a profound break with current logics: entrusting strategic projects to technical experts rather than managers, promoting collaboration between European actors, and fundamentally rethinking our relationship with cloud technology.
The vision of a “Cloud Digital Euro” could provide the unifying framework necessary to mobilize energies and resources at the continental scale, just as the monetary Euro transformed the European economic landscape.
We are at a crossroads. Either we continue to delude ourselves by funding projects doomed to failure because they are poorly designed from the start, or we finally realize that the cloud is primarily a software issue, and that this is where we must concentrate our collective efforts.
Europe has all the assets to succeed: brilliant engineers, innovative companies, a market of 450 million users, and strong values regarding data protection and privacy. What we lack is not talent, but the common vision and political courage to make the choices that are needed at the European level.
It is time to stop being spectators of our own missed revolution and to take back control of our collective digital destiny. The future of European sovereignty depends on it. Not that of an isolated country facing American giants, but that of a united continent, aware of its potential strength and determined to fully exploit it.