<\/p>\n\n\n\n
The Morandi Bridge in Genoa, Italy, was a global symbol of Italy\u2019s leadership in concrete infrastructure, with thousands of viaducts, tunnels and bridges worldwide based on Italian concrete design. But when the bridge collapsed in 2018, it was likened to the collapse of a national myth<\/a>. An infrastructure is not just a void filled with some rubble or cement \u2013 when we design, repair, use and navigate infrastructures, we simultaneously create their myths of greatness and hide their decay.<\/p>\n\n\n\n Throughout history, we have seen time and again how infrastructures have been created through controversy and negotiation; and they are often the result of one people\u2019s dominance over another. It\u2019s not something we usually think about when we move around them. Not until the day they collapse around us.<\/p>\n\n\n\n Stargate, X-files, supercomputers and the cure for cancer<\/strong><\/p>\n\n\n\n Recently, US President Donald Trump, along with some of the biggest US tech executives \u2013 OpenAI CEO Sam Altman, SoftBank CEO Masayoshi Son and Oracle Chairman Larry Ellison \u2013 announced the \u201clargest AI infrastructure project in history\u201d.<\/a> \u201cStargate\u201d, as they have chosen to call the project, has the same name as Microsoft and OpenAI\u2019s supercomputer project<\/a>, but it actually also shares the name with a secret military intelligence program<\/a> that ran from 1977 to 1995, with the mysterious purpose of investigating psychic intelligence methods.<\/p>\n\n\n\n Unlike the old enigmatic military program, Trump\u2019s AI infrastructure project is much more concrete. The US and the country\u2019s biggest tech companies intend to invest $500 billion in developing \u201cthe physical and virtual infrastructure to power the next generation of artificial intelligence\u201d<\/a>. Arm, Microsoft, Nvidia, Oracle and OpenAI are the technology partners in the project, which is primarily about developing data centers to support the development of AI.<\/p>\n\n\n\n It all sounds very technical, but AI infrastructures like this are basically like the Morandi Bridge. They are not just technical details – data centers, billions, chips, digits and cables. They are built on dreams, and not least myths about power and dominance. Jobs, economy, health, military and security were just some of the areas that this new AI infrastructure will facilitate, Trump and his cronies told the press conference. It will create 100,000 new jobs, and like a magic pill, it will also cure cancer, OpenAI’s CEO Sam Altman said. Stargate, can solve pretty much all of America’s problems – but above all it is a national infrastructure that will “Make America Great Again”. <\/p>\n\n\n\n In the same breath, Trump repealed the previous President Biden’s executive order from 2023 on managing AI risks.<\/p>\n\n\n\n The European Infrastructure Dream<\/strong><\/p>\n\n\n\n The EU has also long had an infrastructure dream. This is also about a territory, but as I have described it in my latest book Data Ethics of Power (published with Edward Elgar) <\/a>from 2021, the European infrastructure dream is quite different from the American one.<\/p>\n\n\n\n The European union was established after the Second World War as a collaboration on peaceful solutions between European countries. First and foremost, it resulted in economic cooperation between the countries, but it later developed into a political union around, for example, foreign policy, migration and security. Initially, the European infrastructure projects therefore focused on ensuring the very operation of social and economic cooperation. For example, the trans-European transport network (TEN-T) was created with the aim of \u201cstrengthening social, economic and territorial cohesion in the EU\u201d.<\/p>\n\n\n\n Creating infrastructure in the EU was therefore from the beginning a strategic political effort, which resulted in various infrastructure \u201cpractices\u201d \u2013 such as design standards, construction, investment and regulation targeted at the development of a European infrastructure. The first thing to do was therefore to implement and develop the EU\u2019s physical transport infrastructure, which today consists of over 217,000 km of railways, 77,000 km of motorways, 42,000 km of motorways, 42,000 km of inland waterways, 329 major seaports and 325 airports.<\/p>\n\n\n\n In the 2010s, the EU began to dream of expanding the physical infrastructure with a digital infrastructure. Strategies and policies were launched, such as investments in the technical parts of a digital infrastructure for the European single market in particular. <\/p>\n\n\n\n But then something interesting happened. With the internet, the borders around regional and national territories had exploded. Data flowed between countries and in and out of the EU. The internet\u2019s infrastructure knew no borders, and national or regional digital infrastructure projects and dreams now concerned the entire world. The EU\u2019s infrastructure suddenly became very \u201cgeo-political\u201d.<\/p>\n\n\n\n For example, it created a sensation around the world when the EU made its mark in 2012 with its proposal for a European data protection reform. The proposal to protect EU citizens\u2019 data in the global digital infrastructure was so significant that an American diplomat even spoke of a transatlantic \u201ctrade war\u201d.<\/a> After all, data was the oil of the US big data economy, and it was also what drove the “surveillance industrial complex”<\/a> that we have since got to know very well thanks to the PRISM revelations<\/a>.<\/p>\n\n\n\n A Different Kind of Competition<\/strong> <\/p>\n\n\n\n The European \u201cthird way\u201d \u2013 a term with reference to the geopolitical tech positions of the EU, the US and China \u2013 was only really entrenched in the late 2010s with the EU\u2019s AI strategy and policies<\/a>. From the very beginning, the EU\u2019s \u201cAI agenda\u201d took the form of a geo-political positioning with an emphasis on the development and implementation of \u201cethical\u201d and \u201cTrustworthy\u201d AI. We were two Danes in the EU\u2019s AI High Expert Level Group<\/a>, which was established in 2018 to develop the EU\u2019s ethical guidelines for artificial intelligence. We developed, among other things, seven ethical requirements for AI: Human agency and oversight; technical robustness and security; Privacy and data management; Transparency; Diversity, non-discrimination and fairness; Social and environmental well-being; and Accountability. Requirements that were later transferred to the EU\u2019s AI Act launched in 2021 and adopted last year.<\/p>\n\n\n\n The whole world began to look towards the EU\u2019s risk-based approach to AI and the infrastructures that come with it \u2013 including the US. Now OpenAI\u2019s Sam Altman and Mark Zuckerberg were suddenly to stand tall in the US Senate. Altman called for AI legislation, and Zuckerberg apologized to families of victims of sexual exploitation and harassment on social media. In 2022, the US launched \u201cThe AI Bill of Rights\u201d<\/a> \u2013 a set of guidelines for the responsible design and use of artificial intelligence \u2013 and in 2023, the Biden Administration rushed the AI Executive Order on Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence<\/a>. This should ideally happen before the EU\u2019s slow democratic negotiation of AI legislation was finished, because back then it was all a matter of coming first in the ethical global \u201cAI race\u201d.<\/p>\n\n\n\n Now, this was a fantastic competition to participate in. Because some really good collaborations were also established on that basis. The EU worked with what in diplomatic circles is referred to as \u201clikeminded\u201d countries around the world \u2013 that is, countries built on the same values of democracy and human rights as the European Union. They worked on ethical principles for AI and handling AI risks in the UN, OECD and G7, and with countries such as Canada, India, Japan and the USA.<\/p>\n\n\n\n At the time, I was working on a large EU foreign policy project, InTouchAI.eu<\/a>, which supported the EU Commission\u2019s work with the EU\u2019s \u201chuman-centric\u201d approach to AI. In a small group, we developed, for example, the EU and US Trade and Technology Council\u2019s \u201cJoint Roadmap for Trustworthy AI and Risk Management\u201d<\/a> where, without much complication, we agreed on a common starting point:<\/p>\n\n\n\n \u201cto advance our shared interest in supporting international standardization efforts and promoting trustworthy AI on the basis of a shared dedication to democratic values and human rights\u201d.<\/p>\n\n\n\n “Totalitarian Infrastructures of Power”<\/strong><\/p>\n\n\n\n We were busy building bridges around the world. We promoted the EU\u2019s AI legislation and human-centric approach at the Dubai World Expo, the Paris Peace Forum, COP27 in Egypt, the AI Alliance Assemblies of two EU presidencies, and several other initiatives. When a French journalist caught up with me in 2023 after a panel discussion with a British parliamentarian and a US commissioner, I was optimistic. At WebSummit, Europe\u2019s most hyped tech conference, we were actually discussing AI legislation in the EU and the US. I highlighted the prospects for a more responsible global direction, but the journalist focused only on my concerns:<\/p>\n\n\n\n \u201cGry Hasselbalch, a Danish academic who advises the EU on the controversial technology, argued that the West was also in danger of creating \u201ctotalitarian infrastructures\u201d. <\/strong>\u201cI see that as a huge threat, no matter the benefits,\u201d she told AFP\u201d. <\/a><\/p>\n\n\n\n I was annoyed that my optimism did not make it into the article. But today I must admit that I was probably way too gullible in terms of how quickly good intentions can change once the infrastructure is already in place \u2013 and it was in place, but it was not an infrastructure built on the European dream. Our digital infrastructure was never our own.<\/p>\n\n\n\n American infrastructure in Europe<\/strong><\/p>\n\n\n\n And this year, things suddenly went very fast. It actually only took a change of government and a few days of a week. First, a new president replaced another president’s executive order with a new executive order: REMOVING BARRIERS TO AMERICAN LEADERSHIP IN ARTIFICIAL INTELLIGENCE.<\/a> Then the new president appeared on the screen with his tech buddies, and during a very short televised press conference, the new, slightly safer bridge we were building collapsed, and the sophisticated “infrastructurering” of not only our space, but also time<\/a> that started decades ago suddenly became very visible again.<\/p>\n\n\n\n Because while we have been busy dreaming of a slightly safer bridge, it was actually the American dream that we were implementing. There were all the obvious virtual infrastructures: Facebook became the infrastructure for our social relationships, Google for everyday life and school, Microsoft for the workplace. And then there was the physical infrastructure itself for the digital services. In 2024, Microsoft announced an investment of \u20ac3.2 billion in Germany to double its AI infrastructure and cloud computing capacity, they did much the same in Sweden. Meta has also invested in data centers in Europe, including in Denmark, and Alphabet (Google) has repeatedly announced large investments in expanding its data center infrastructure in Europe. And here, as one of the most digitalized countries in Europe, Denmark, which will take on the EU Presidency in July 2025 naturally plays an important role, as one enthusiastic consulting firm<\/a> puts it:<\/p>\n\n\n\n \u201cDenmark is a gateway to Europe for global network infrastructure, as shown by the case study of Google. The company is now facilitating even greater EU-wide connectivity via Denmark. It has done so as part of a wider infrastructure programme which \u2013 specifically over the period 2018 to 2020\u201d.<\/p>\n\n\n\n Socio-technical Infrastructures of Power<\/strong><\/p>\n\n\n\n Welcome to the future. We gave it away ourselves. In Data Ethics of Power, I described the development of two different kinds of \u201cpower infrastructures\u201d:<\/p>\n\n\n\n Big Data Socio-Technical Infrastructures of Power (BDSTIs):<\/p>\n\n\n\n These are the infrastructures that primarily maintain and amplify the power dynamics of a big data economy and society. They are based on dreams of the possibilities of an unlimited big data resource.<\/p>\n\n\n\n