Freeing Europe from Cloud Giants 

Beyond rogue scenarios and hype, the AI boom is accelerating a deeper structural shift: the migration of the global economy to the cloud. So far, the EU has mostly tried to rein in Big Tech through regulation, but this reactive approach has fallen short. The recent push for tech sovereignty can only succeed if it starts with democratic choice and a shared vision for social and technological progress.

Alessio Giussani: Your book The Rulers is primarily about three companies: Google, Amazon, and Microsoft. What sets them apart from other big corporations?

Cecilia Rikap: In the current debate, there’s a lot of noise about what’s going on with AI. Everybody, of course, has the right to talk about AI, particularly because decisions are made behind closed doors, and we lack a space to democratically discuss the ways forward. But instead of talking about the roots of corporate power and how it dictates technological change and profits from it, the conversation is often about two types of agents – and I’m playing with the word “agent” here.

One is the AI agent. People marvel at the things that AI can do, and instead of trying to understand who is developing this technology and for what purposes, they get caught up in this rhetoric that AI is either wonderful or horrible, but is extremely powerful either way. Others, who are more concerned with economic inequalities – as I also am – focus on another type of agent: the CEOs and the founders of these companies, particularly the ones who side with the far right, like Elon Musk or Alex Karp. I’m not saying that this approach is not relevant. But if we truly want to understand the roots of corporate power today, and therefore act upon it, dismantle it, and create alternatives, we cannot miss the forest for the tree.

These far-right characters shout very loudly and sometimes distract us from what’s really going on; this is what happens often with someone like Elon Musk. I’m not saying his companies don’t matter or that we shouldn’t study them, but in my research I don’t want to single out companies arbitrarily. Instead, I try to offer a more comprehensive view of corporate power. Therefore, I start by analysing the largest companies by market capitalisation, and then try to understand what explains their power.

I identify two groups: one is the group of big oil companies, and more broadly, companies that get their advantage by systematically extracting and monetising nature in different forms. The other group, which is the largest one and has been growing fast and overtaking some of the old giants, is made up of companies whose advantage comes from systematically capturing intangibles.

Corporate power has always been built, to some degree at least, on capturing and making sense of intangibles like information and knowledge, organising them, and putting them at the service of production processes. Think, for example, of the R&D laboratory of IBM, or the marketing department of Coca-Cola. These companies have been intellectual monopolies for decades, capturing intangibles like science and knowledge (in the case of IBM) or narratives and brand-building (Coca-Cola).

As intellectual monopolies, they mobilise their intangible assets to exercise economic power. Start-ups, for example, control intangibles, but usually they are not used for capturing economic value. An intellectual monopoly puts different forms of knowledge, narratives, and data at the service of expanding its control of the economy. This control can take the form of a chokepoint if it occurs on a specific market, or of a panopticon if it gives the ability to monitor all information created and exchanged within a network, enabling the leading corporation to plan, organise, and control beyond ownership.

Based on your analysis, a company like NVIDIA – which designs the most advanced AI chips and is the largest in the world by market capitalisation – does not qualify as a “ruler”. Why?

Nvidia has a clear chokepoint because it controls a specific market and is certainly an intellectual monopoly. But beyond that, it doesn’t have the capacity to control a broad part of the global economy. Its power is focused on semiconductor design, whereas other companies have a panopticon view, often because they sell directly across multiple markets. When you are at the supermarket, you think there are so many brands, but then you start turning the packaging, and you see Unilever everywhere. This gives a company like Unilever a panopticon: broad oversight across different parts of the economy. So all these companies mobilise intangibles to control portions of our lives, but these end up being contained, limited. Amazon, Microsoft, and Google are the only ones that combine multiple chokepoints and the largest panopticon of Western capitalism.

Many powerful actors are involved in producing an AI model. NVIDIA’s semiconductors are manufactured in Taiwan at TSMC, which I also classify as an intellectual monopoly with a chokepoint. So I’m not saying a company like Nvidia or TSMC isn’t powerful. But even the semiconductors from Nvidia end up entering the cloud of Amazon, Microsoft, and Google. Everywhere we look, we see these three companies controlling the many pieces and actors that then, seen together, become the digital stack.

Take academic research on AI, for example. Amazon, Microsoft, and Google co-produce around 90 per cent of their research with thousands of organisations. This isn’t exclusive to them: Meta, Nvidia, and companies from other sectors do it too. But when it comes to AI, Google and Microsoft are those with the largest capacity to control universities and public research organisations developing frontier research on AI. Moreover, they control the largest number of other companies, including start-ups, with their venture capital money and through the cloud.

Nvidia is expanding its profits so much that it has leverage to expand beyond its chokepoint. It is now trying to get its own cloud by investing a lot of money – also in partnership with other leading companies – to basically offer its semiconductors as a service on independent data centres that are not those of Amazon, Microsoft, and Google. Of course, this makes total sense because Nvidia doesn’t want to depend on selling to other big players. To some extent, what NVIDIA is trying to do is build its own cloud, and for me, this shows how important the cloud is.

What exactly is the cloud? And why does it bring so much power to the companies that control it?

Let me start with an example everyone knows: Netflix. It looks like a simple streaming app where we can watch whatever we want. Behind Netflix, there are thousands of digital services, which means that you have millions of lines of code that do a lot of different things. Not even a company as large as Netflix writes every line of code that makes the platform exist for all of us. Why? Because there are a lot of services, a lot of ready-made pieces of software for solving specific and repetitive issues that can be purchased. Where do you buy that? In the past, organisations used to go to a software provider and ask for software for their needs. It was more tailor-made, but it was also more expensive. Now, they rent it in the cloud.

The cloud is like a supermarket for digital technologies. It’s not just infrastructure offered as a service – that is, the data centres where data is stored and processed. You can get whatever you want in the cloud: software, a model, data, a platform, infrastructure, and more. From each of these clouds – which are run by Amazon, Microsoft and Google – you have thousands of options that you can purchase as a service.

That also means you never truly get access to the code. You don’t download the service to your computer, but you can use it while you are connected to the cloud. And this is where the cloud starts being something different. It’s not just the supermarket of technologies. It is also the production sphere, the factory, the place where data is arranged and organised, the software is produced, the architecture of the platforms is designed. Everything happens inside the cloud, and when we consume the services, we’re also inside the cloud. It’s as if I were also cooking and eating my meal inside the supermarket where I bought the ingredients. The production, the exchange, the consumption: it all happens there.

This differs from the power these companies already had over individuals through social media. Organisations of every size and every type, from small firms to large companies like Netflix, Nestlé, or Coca-Cola, and even governments, migrate to the cloud. In fact, the entire global economy is migrating to the cloud, which means that companies that control it get the capacity to oversee, predict, and anticipate where the economy is going; a level of foresight that no single institution, government, or organisation had in the past.

The cloud and intellectual monopolies predate the explosion of AI hype with the public release of ChatGPT in 2022. What does AI fundamentally change in the cloud’s role and the power relations it generates?

The AI companies that we now associate with the AI boom or hype, like OpenAI and Anthropic – but also DeepMind, which was acquired by Google in 2014 – are tightly connected to the cloud giants in multiple ways, including venture capital money. Since the release of ChatGPT, these AI companies have required a growing amount of processing power, which has also driven the data centre boom. At the same time, we have organisations around the world that feel that they must adopt AI, that they cannot be left out. I call this AI FOMO. And what is the safest, and also the most seamless, way to adopt AI? Well, the cloud.

AI is like a Trojan horse, a marketing strategy for the cloud. It accelerates migration to the cloud. Through Amazon, Microsoft, or Google Cloud, you automatically integrate the AI models into all the rest of the digital services you use. And you might not trust Anthropic, but Microsoft? Microsoft has been there for many years, so these companies have built a reputation as trustworthy partners, especially for other leading firms. Some companies have contracts directly with OpenAI or with Anthropic, but more and more we see that organisations prefer to go to the supermarket, where today they might use Claude, tomorrow ChatGPT, and the day after tomorrow they may use China’s DeepSeek, because even DeepSeek is available as a service on the cloud of Amazon, Microsoft, and Google.

Meanwhile, regulators and competition authorities focus on AI and whether the AI market will be competitive, or what to do to make it competitive. But AI exists within an ecosystem, and the ones that control the ecosystem beyond ownership are the cloud giants.

Controlling beyond ownership is also linked to planning and delivering innovation. How do you understand that concept?

We tend to think that innovation is a precondition for economic growth, and that growth, if we have the right policies, will lead to prosperity and a better life for everyone. Private companies are seen as the innovators, so governments are expected to cut red tape so we can achieve prosperity. Yet societies are not becoming more prosperous.

The classical understanding of innovation would define it as a new process or product put on the market. An invention might be the discovery of a new drug or treatment, but until it’s approved and sold on the market, it’s not an innovation. But this definition neglects that every innovation is produced collectively by many, and then monetised by the company that captures that co-created knowledge. Between 2012 and 2021, for example, Microsoft produced over 15,000 scientific publications. Almost all of them had at least one co-author who worked for another organisation, not Microsoft. During the same period, Microsoft was granted 35,000 patents. It shared ownership of just 1 per cent of those patents.

This goes to show that these corporations are not the only innovators. When companies like Microsoft, Amazon, Google, and others work with public universities and research institutes or invest in startups, they do so to get access to the knowledge that is being produced, to steer the trajectory of that knowledge, and then recombine all the pieces of that huge puzzle and monetise all the new recombinations. In a way, they are the big appropriators. So we shouldn’t cope with them under the promise that their innovations will contribute to a better life. In fact, they don’t, and these are not just “their” innovations, but the product of collective creative processes involving many others. These companies are indeed producing a lot of frontier research on AI, but never alone.

Besides, innovation today should be understood as knowledge in all its forms. This includes narratives and data, and how all this gets combined in the melting pot of creativity that the intellectual monopolies control beyond ownership. Corporations craft narratives about what we think AI is and how it should be regulated, and push us to use AI as a method of inventing, sidelining other means of producing and creating. I call this process epistemic totalitarianism.

Europe seems to have only awakened to the dangers of this system since Trump’s return to the White House in 2024 and his unholy alliance with tech bros. But who’s really in charge? Who’s driving whom?

During the Cold War, the Department of Defence set the trajectory and orientation of technological change. The Pentagon did this by investing strategically as well as coordinating and planning the big technological innovations of the Cold War period. After the Cold War, there was a sense of, “Why should we continue doing this the same way? We won.” Of course, it doesn’t mean that the Pentagon stopped investing in technology in the 1990s and 2000s, but it was no longer steering and planning technological development in the same way. Now, big tech companies have increasingly become the big planners of science and technology, particularly of AI. The AI that we have is the AI that they want. It’s an AI that can be resold again and again to countless actors, making it cheaper to produce and expanding their control over our societies.

Europe has been missing this story for decades. It thought that the world was operating as a common market, and that we should promote competition because everyone was on the same page. Of course, Europe knew that China wasn’t on that page and that China was planning. But what was harder to see was that leading corporations, in particular those from the US, were also planning and controlling beyond ownership.

The US government, once it realised around a decade ago that this was going on, decided to take action, and the already solid relationship between big tech and the Department of Defence has grown stronger and stronger since then. The most important change in this dynamic is that big tech – unlike previous military subcontractors – does not depend on government money. A key difference between a company like Palantir and a company like Microsoft is that Microsoft doesn’t depend on funding from the US government and its allies to be profitable, whereas Palantir does. This doesn’t mean Palantir will always obey the White House rule and bend the knee, but it is in its best interest to keep the government happy. For big tech companies, it’s not about getting money from the government; it’s about getting political help. They need the US government to stop places like Europe from regulating them.

What do they want exactly? They seem to be asking to pull the brake on AI development, while promising that AI could bring immense benefits to humanity and pushing on the accelerator.

There are two prevalent discourses in Silicon Valley, which correspond to two different ways of trying to control the state. One is that democracy, innovation, and freedom cannot go together. It posits that innovation and freedom should be prioritised over democracy. For those pushing this view, the state has failed and needs to be destroyed; technology by itself will bring a better world and deliver progress in every respect – to those who can pay for it, of course. These are the views of people like Alex Karp, Elon Musk, and Peter Thiel. But we also have the discourse of Satya Nadella and Sundar Pichai, the CEOs of Microsoft and Google, who say almost every day that their technologies will deliver promised economic growth while helping governments manage public administration, schools, hospitals, and borders.

This second discourse is more subtle. I describe it as corporate diplomacy because it’s a more strategic way to convince governments – even progressive ones. And when it comes to regulations, of course, these companies will respond differently: they will pretend to be much more open to oversight, but the question is which type of regulation and who designs it. They say that the technology moves so fast that it’s impossible for government regulation to be updated as it should be, and so they push for self-regulation, or regulation done by a third-party actor that goes inside the company and works together with its employees, with the information that the company wants to share. And they agree on regulations that blame tech users but never (themselves as) producers.

So, even though it might seem like they’re saying something different, in the end, they are all pushing for a certain type of regulation that blinds governments and society more and creates a shield from external oversight.

Regulation has been the approach driving policymaking in Brussels for a long time. Now there has been a shift towards competitiveness and a new buzzword: tech sovereignty. You’ve been working with this concept well before it became so popular. What does it mean?

Tech sovereignty can mean three different things. It can mean techno-nationalism, which consists in giving all the contracts to European firms. However, European companies like Siemens or SAP are also intellectual monopolies. And they are private companies, so their interest is to make money, not to grant options, choices, and openness to our societies in terms of how technology is produced or for what purposes.

Another interpretation, prevalent today at the EU level, sees tech sovereignty as making things happen in Europe, and it is as problematic as the first, if not more. The European Commission’s Tech sovereignty package should be renamed the tech competitiveness package. It aims to achieve competitiveness and economic growth through or with the excuse of sovereignty. It’s good marketing to call it sovereignty, but it’s essentially about attracting more capital to Europe, growing GDP, and reducing the feeling that we are lagging behind vis-à-vis the US. As it is designed, the package allows a company like Microsoft to fulfil all the criteria [of Europeanness] if it stores data in European data centres, and if its subsidiaries selling its services are based in Europe. So, from a legal point of view, these companies will be European entities, even if they are part of the Microsoft multinational corporation.

Clearly, these two ways of thinking about sovereignty are very short-sighted. Sovereignty, when it comes to tech, should be connected to governance, democracy, and choice. It is about being able to decide which technologies we want, which ones we don’t want, and for what purposes. Who is going to develop them? To what extent are we going to develop them, given their ecological footprint? What data are we going to use? How will we recognise the collective work that goes into developing technology? How are we going to make sure the effects don’t harm workers?

Ultimately, sovereignty is about who decides. It’s about democratising the political power of who is deciding about the present and future of humanity, because these technologies are powerful and ubiquitous, and the technologies that we are getting today are decided behind closed doors by a few actors.

If you were a progressive politician, government, or policymaker who wants to work towards this more expansive notion of tech sovereignty, what kind of coalitions and constituencies would you try to involve in the process?

The “we” needs to include everyone who identifies that this is a problem and wants to build a vision of digital sovereignty that is about collectively co-creating and sharing knowledge.

I think we need to start by decoupling the public sector from these companies. We need to make sure the US government and the companies can't switch off the states participating in building an alternative. We can start small, for example with education, which is a critical sector, because kids at school are receiving a Google Suite package and studying with generative AI. We’ve seen the effect it has even on adults: it reduces people’s capacity to think critically. We, as adults, have learned to maintain some balance, but kids often don’t have that capacity. Other critical sectors are healthcare and the public service media.

It’s hard, but it will not require all the money that the companies are investing because we are going to start smaller, from the capabilities that already exist. And Europe has a lot of those. A significant part of these companies’ advantage comes from research that was done and is still being done in Europe by European universities, European companies, and so on.

There’s a growing movement against data centres. Should they be part of the conversation? And can we really build a democratic and sovereign tech if we put a pause on them altogether?

Including the communities that are fighting against data centres is critical because if we are building a public alternative, we will need infrastructure. The question is, where are we going to build these data centres? To what extent do we need them?

There is the idea that hyperscale data centres are inevitable, but only because we’ve accepted the narrative from cloud giants and AI companies that what we need is the largest model with more parameters, even though that is less rigorous than the machine learning models developed to detect or do specific tasks. But of course, those models require more human effort and less energy. So, we also need to change the AI paradigm in this process.

Tech CEOs and politicians in the US mention China “catching up” day in and day out as the ultimate reason to push the accelerator and push back against meaningful regulation. Is that narrative justified by reality?

When you look at the level of investment in the digital infrastructure of Amazon, Microsoft, and Google compared to Alibaba, Tencent, or Huawei, you can clearly see that China is playing a different game. It is trying to expand the adoption of its models as much as possible, even if they are not at the frontier. If it had political interest in doing so, China could channel all its efforts into one big team producing the largest AI model you’ve ever seen, and it has the resources and people to do that. But that’s not the game that China is playing.

Nor is Beijing just playing the AI game. China has been focusing on so many industries; it’s trying to lead everywhere, while the US is focused on AI, and Europe is stuck in between because it’s not thinking about what Europeans' problems are today. What’s the priority? What should we be focusing on?

I like to live in Europe. The quality of life here is way better than in the US, and we often forget this. We forget about all the things we still get in Europe that are much rarer in other parts of the world, and I think we need to embrace that much more. It’s the time to build digital commons. It’s the time to build global public goods. It’s time to solve critical problems instead of being blinded by the latest iteration of an AI model that clearly brings more problems than solutions. And again, I’m not saying we should solve all problems without technology, but we should first map the problems, who is being affected by them, and how, and then identify what type of solutions is needed. In some cases, it might be technology; in other cases, it might just be redistributing wealth and decision-making power. It's ultimately about bold political choices, and I hope to see more of those coming from Europe.

The content of this interview has been edited and rearranged for length and clarity.