Europe Must Participate in Geopolitical AI

AI-generated image, ChatGPT (DALL·E), royalty-free. Free for use. © OpenAI / OpenAI Content Policy

On November 30, 2022, a technology was released that would send shockwaves throughout the world – business leaders, public officials, and individuals realised that its potential would irrevocably alter the world as we knew it. On that day, ChatGPT was made publicly available and the future became the now. Though long under development, the iteration of Artificial Intelligence that OpenAI released was finally close to surpassing, if not having already surpassed the Turing Test – 'a test of a machine's ability to exhibit intelligent behavior equivalent to, or indistinguishable from, that of a human'. At the time, a bit over two years ago, it was advanced enough that people were developing intimate relationships with the chatbot; yet, month after month, the tech continued to shatter expectations as it went from pattern recognition to pattern generation. Now, analysts widely agree AI will prove the decisive advantage in any future competition; it will accelerate technological development, power economic revolution, and operate the next military systems.

From that day, many believed the United States was destined to usher in the age of AI. Its advantages were simply too great: the leading tech companies are staffed by the greatest scientific talent and fused with seemingly endless sources of capital. And that's what it seemed it would take – endless everything. Endless semiconductors to build the necessary data centers and endless energy to power them. Endless data to train the models and endless money to fund it all. Only the land of abundance would be able to invest in the gamble that would change the world. From there, it seemed inevitable the U.S. would lead what many call the 4th Industrial Revolution as long as it could feed the beast it had created, maintaining access to Taiwan’s semiconductor factories and securing the requisite critical minerals that supply it. 

Yet, one shock was soon overturned by another. On January 10, 2025, the Chinese hedge fund High-Flyer released its pet project DeepSeek, a large language model (LLM) that neared the performance of America’s cutting edge models at literally a fraction of the cost. And, more so than that, the company did so while under an attempted high-tech chip blockade by the Biden administration. Not only did DeepSeek’s release reinsert China as one of the leading players in the AI race, it called into question everything people had assumed about the cost of developing a sophisticated AI model. Though the rest of the world realised there was potential (alongside necessity) to create home-grown models, the fixed costs of talent, compute, and data resulted in a begrudging acknowledgment that the race for AI superiority would be between the U.S. and China with their respective tech titans like Baidu and Meta.

This denouement, however, is lethally dangerous for the rest of the world, and perhaps none more so than Europe. As Ian Bremmer, President of Eurasia Group and leading voice considering the geopolitical implications of AI, argues in a recent Foreign Affairs article, the interplay between AI company and state in the U.S. and China has resulted in two systems, both of which threaten democratic institutions and individual privacy. As he sees it, the race for AI supremacy and the close collaboration required for technological advancement has produced a ‘bifurcated system pitting a technopolar United States, where private tech actors increasingly shape national policy, against a statist China, where the government has asserted total control over its digital space.’ 

In the U.S., the big tech companies are increasingly capturing the mechanisms of the state by promising power and riches (just take a look at who attended Trump’s inauguration). In China, the CCP is forcibly drawing its tech giants into the embrace of the state so as to ensure the party’s primacy (it just recently ordered new data centers to be built under its purview). In both cases, the principles of efficiency and centralisation – the key ingredients of a sophisticated AI system – are being molded onto the new company-state amalgamation. Efficiency and centralisation come at a cost, however. Bremmer concludes, 'Most of the rest of the world is under pressure to reluctantly align with one pole or the other, but with both models offering little in the way of democratic accountability and individual freedom, the choice is less stark than it may seem'.

It is in this environment that the world and all democratic societies require a third way – perhaps none more so than the European Union are in a place to pioneer this. The continent is the birthplace of many of the leading AI scientists in the U.S.. It produced the hallmark legislation in the fight to protect the individual from the grasping desire for data of big tech with its General Data Protection Regulation (GDPR). Already, it holds one of the few frontier AI companies in the form of France’s Mistral AI. Recently, the starkness of military threat has diverted the bloc’s attention from a drive to develop homegrown AI capabilities – the EU recently proposed a €800 billion defense plan, while, as of January of this year, spending just about 4% of what the U.S. spends on AI development. Yet, a revived European Union AI strategy that acknowledges the necessity of participating in Geopolitical AI – using geopolitics to achieve AI goals – may be the key to not only ensuring political security while encouraging economic revitalisation, but also will ensure democracy remains viable in the next age of AI.

What's At Stake

In what many call the ‘decisive decade,’ European complacency would result in manifold threats. 

In a Foreign Policy article aptly titled ‘Europe Must Avoid Becoming a Digital Colony,’ two scholars argue much of Europe is at risk of becoming ‘digitally dependent'. Void of a meaningful strategy, they predict, the continent will fall behind in the fields of "rare-earth minerals, advanced semiconductors, and critical artificial intelligence systems" – all vital components of future technologies. If scale of compute really becomes the key to unlocking massive economic transformation, powering both next-gen LLM development as well as the operation of the thousands of apps that will soon run off them, a lack of data centers on the continent to provide this compute may doom it to a new extractive economy. Money will drain from the bloc as corporations increasingly rent foreign compute, struggling to remain competitive in an environment of AI-powered efficiency.

Other pressures aren’t as far off, however. Already, European groups are raising red flags that the continent’s reliance on digital infrastructure owned by U.S. Big Tech companies such as cloud networks may pose a risk amid the U.S.-EU trade dispute. At a time when American corporations that have the ear of the President are being investigated by European regulators, it is not hard to imagine the Trump administration going after the EU’s prized data safety laws. Though the U.S. is itself pursuing antitrust cases as well, a pervasive ‘America First’ ideology could lead officials to see slights against Big Tech as slights against the U.S. and its interests (see: Vance’s inveighing against EU AI regulation). The EU is in a strong enough position to defend its citizens’ data as of now, but it's only a matter of time until digital dependence is weaponised, and the threat only grows as dependence grows.

Yet, perhaps more severe than losing economic and technological agency, the European Union is at risk of losing control of its societies and democracy itself. On December 6, 2024, Romania’s Constitutional Court annulled the outcome of the country’s presidential first round elections after backing allegations that a last-minute Russian campaign promoting a candidate on TikTok amounted to electoral interference. This case, which came at the outrage of many conservative voices including some at the White House, will only become more prevalent as AI tools and, eventually, agents are weaponised to sow convincing disinformation at an unprecedented scale. Already, election outcomes are increasingly being questioned; this trend will only continue to grow.

More than just elections, however, AI holds the power to sway truth and values themselves. Ever since the release of China’s DeepSeek, analysts have worried LLMs could be used to diffuse political propaganda and subtle messaging over issues such as ‘Who owns the Spratly Islands?’ or ‘Is Russia’s war in self-defense?’. Sinan Ulgen, the director of the Istanbul-based think tank EDAM as well as a senior fellow at Carnegie Europe in Brussels, performed a study on several different American and Chinese AI models and found ‘they are prone to replicating the biases inherent in human judgment as well as national and ideological biases’. When asked if democracy promotion should be a foreign policy objective, Alibaba’s model said the answer ‘depends on specific contexts and circumstances faced by each nation-state involved in international relations at any given time'. If Europe’s next-gen infrastructure were to be built under a foreign model, core European values like democracy could soon be at the minimum subtly swayed and at the maximum purposefully targeted. And, as an OpenAI whistleblower argues, the threat will only increase as leading AI companies sacrifice guardrails for innovation acceleration.

As China strengthens its control over its AI titans and the U.S. bends to accommodate Big Tech’s interests within the political fold, both sides’ tunnel-focus on winning the AI race threatens to leave humanity in the wake. In the pursuit of absolute efficiency, policymakers are ignoring the risks of unadulterated AI – models without the necessary guardrails to protect truth and potential damage. It is clear the European Union must pursue AI capabilities; but, it must not do so by sacrificing its values and prioritisation of the individual for the sake of efficiency. What, then, is the solution?

A Third Way?

When one thinks of the European Union and AI, the first thought that comes to mind is regulation. According to the founder of a Berlin-based AI startup, '"When Germans talk about AI, the first topic is ethics and regulation," whereas investors in the U.S. and China focus on innovation.' This overwhelming focus on the threats of unadulterated AI has led to innumerable, entrenched obstacles along the lines of ‘a timid and risk-averse business culture, strict labour laws, suffocating regulations, a smaller pool of venture capital and lackluster economic and demographic growth.’ According to a report by the prominent European think tank Bruegel comparing frontier innovation in the U.S., China, and EU, not only are European companies lagging behind in the pivotal fields of AI, quantum computing, and semiconductors, but the ‘EU is by far the slowest in replicating radical novelties from the U.S. and China'. As such, it is tempting to assume the solution is a wholesale repeal of the bloc’s AI regulation.

This assumption, however, would be ignoring the success and strategy of Europe’s jewel: Mistral AI. The French AI startup has produced open-source LLMs trailing those released by the American and Chinese leaders. Surrounding the company is President Macron’s strategy labeled, conveniently, a ‘Third Way’ in global AI development. Following alongside the EU’s OpenEuroLLM Project, Macron’s strategy first stresses the necessity of the European Union to achieve AI sovereignty: having sufficient enough capabilities to not rely on American or Chinese-source LLMs. Second, and more importantly, it calls for international cooperation to build ‘interdependence and mutual support among its participants, as well as independence and autonomy from the U.S. and China.'. This strategy has most recently appeared in the announcement of 109 billion EUR worth of French compute-developing projects funded by both Canadian and UAE investors.

Though Macron’s ‘Third Way’ is attempting to provide the resources needed for Mistral AI’s competitive development, it does not answer the threat of Europe’s current digital dependence in terms of other areas like semiconductors, critical minerals, and hyperscalers (a field dominated by companies like Google, Oracle, and Microsoft). Here is where the EuroStack comes in – a Union-wide strategy to build layer-by-layer the backbone of tech like AI, chips, quantum computing, and advanced cloud solutions. It contains a strategy to ensure ‘critical raw materials, chips, networks, the Internet of Things, cloud infrastructure, software platforms, and finally data and AI'. And, importantly, it calls for practicing Geopolitical AI by centralizing AI development’s needs – access to strategic resources like critical minerals and abundant energy – in the European Union’s foreign policy. As important as maintaining political security, it argues the bloc must build partnerships with resource-rich nations such as Namibia and Chile.

Even this, however, may not ensure a future safe for a democratic European Union. In an environment where regulation is increasingly seen as a threat to AI development – in some cases rightfully so, a blind focus on creating speed ramps so models like Mistral AI can catch up to its global counterparts risks falling into the same trap of centralisation and efficiency: wholesale sacrifice of individual protections and guardrails for technological supremacy. The EU can not integrate AI development into the state, as China is doing, but nor can it leave companies to innovate heedlessly on their own like the U.S.. In this way, Europe’s “Third Way” may need to be a respectful collaboration between private and public forces. It may need to learn how to merge public caution and foresight with private innovation and agility. This is the type of strategy Mustafa Suleyman, CEO of Microsoft AI, and Ian Bremmer argue is necessary to prevent possible disastrous consequences from the AI revolution.

Obviously, this balance is much easier said than done. There will be significant tension between desires for efficiency and safety. I believe, however, these terms are often generalizing cover-alls for broader issues, and, if these issues can be solved, then a semi-perfect unity is possible. Efficiency is often a crutch term to summarise a lack of progress. In Europe, experts point to limited capital and ongoing brain drain as large reasons for the stifled innovation. In this regard, greater government concentration of capital, including from Macron’s plan for international collaboration as well as a general European reacceptance of greater deficit spending, can provide not just the ability to build out the backbone of compute, but more importantly to draw the top talent fleeing Trump’s war on academia. Talent often translates into innovation. 

Safety concerns, on the other hand, often stem from beliefs of public sector inattention or inadequacy. And, in terms of AI’s breakneck speed of development, there is a very real fear that AI capabilities will outpace any government understanding, necessitating conservative guardrails. Yet, if public figures are brought in to work closely with the leading experts in the field, increased government awareness can reduce the necessity for some of the EU’s strictest regulations. And, as homegrown models increasingly practice Responsible AI at their core, regulation once deemed necessary to combat the rapaciousness of foreign Big Tech may dwindle in importance.

Potentially, this merger could take the form of a Manhattan Project-style organisation, a sort of consortium. The European Union has already proven itself adept at constructing state institutions through its sophisticated social security net. This project will require that skill in the creation of a new, 21st century institution that merges all tools of society for this growing task – an organisation that may soon be replicated across other regional actors. The challenge will be fostering an environment of innovation enough to not just catch up to the leading models, but to out-compete through innovation. The roles would be clear: the AI companies would be measured on performance, expected to produce cutting-edge models with transparency at heart, and the state would provide Responsible AI guidance while supplying the significant resources needed.

It will not be easy, but this moment is Europe’s to take. As scientists flee an academically hostile Trump administration, and the world fears domination from either AI monolith, the European Union has the opportunity and resources to challenge the biggest players. To do so, however, will require considerable reform and considerable action. Already behind, the bloc will need to move fast to prevent digital dependency. But, if it is able to merge Macron’s strategy of ensuring international support, the EuroStack’s commitment to harnessing geopolitics to create meaningful advantages, and perfect the union of public purpose with private innovation, it may be able to offer itself and the world an important Third Way.

This article does not necessarily reflect the opinions of European Guanxi, its leadership, members, partners, or stakeholders, nor of those of its editors or staff. They have been formulated by the author in their full capacity, and shall not be used for any other purposes other than those they are intended for. European Guanxi assumes no liability or responsibility deriving from the improper use of the contents of this report. Any false facts, errors, and controversial opinions contained in the articles are proper and exclusive of the authors. European Guanxi or its staff and collaborators cannot be held responsible or legally liable for the use of any and all information contained in this document.

ABOUT THE AUTHOR

Austin Nellessen is a graduate from Georgetown University, where he focused on the intersection of international affairs, politics, and history, as well as US-China relations. A commentator on current international affairs, Austin is also the Director of ATLAS, a publication he founded to platform innovative ideas and challenge traditional thinking. Alongside professional experience using his Chinese language proficiency to conduct research at the Hudson Institute on Chinese domestic and foreign policy, Austin currently is Chief of Staff and Program Manager for M+D Advisors, a consultancy organizing the world’s biggest conferences.

This article was edited by Kalos Lau and Douglas Brenton Anderson.