Europe’s Role in the Global AI Landscape

Author

Nikolaj Munch Andersen

Office of Denmark’s Tech Ambassador

Biography

Nikolaj is the Chief AI Advisor at the Ministry of Foreign Affairs of Denmark, working at the Office of Denmark’s Tech Ambassador to assist Denmark in navigating a new geopolitical reality that is increasingly shaped by technology and the large companies that control it. Nikolaj holds a Master of Science degree in Cognitive Science and has a strong technical background in language technology, having coauthored papers accepted to NeurIPS, EMNLP, and Nature Medicine. He has previously served as Data Science Mentor at MIT Critical Data and Affiliate Researcher at AIM, the AI in Medicine Program at Harvard Medical School. Nikolaj writes this piece in a personal capacity. The views expressed in this publication are solely those of the author and should not be attributed to their employer or any affiliated organizations.

Nikolaj writes this piece in a personal capacity. The views expressed in this publication are solely those of the author and should not be attributed to their employer or any affiliated organizations.

Advances in neural network architecture over the past decade have transformed how machines encode, process, and generate text and images. What began as a Google research effort to improve its—to some, infamous—translation service led to the development of the transformer architecture.[i] With the transformer, Google researchers introduced a new way for computers to understand language by looking at all parts of a sentence at once, rather than processing them step by step. This approach allowed models to better grasp context and relationships, making them faster to train and far more capable, laying the foundation for today’s breakthroughs in generative AI.

The methodology was released openly, and soon, a small research lab in Silicon Valley began experimenting with it, scaling the models, and expanding training data.[ii] The results signaled a clear shift in what these systems could do given the right computing budget and internet-scale text data.

And here we are, in 2026. That small AI lab is no longer small; in fact, it is now the most valuable private company in the world. The uptake of generative AI has been unprecedented, far outpacing earlier technological shifts, such as social media, streaming, and smartphones.

The US economy has been significantly driven—some even say propped up—by technology giants whose main focus is now on AI systems and the infrastructure that sustains them.[iii] By many, especially those who benefit most from this global uptake, this wave of AI is described as “the new electricity” or a “general-purpose technology” reaching across sectors and disciplines.[iv] Although this may be useful as a metaphor for diffusion and economic potential, these labels tend to oversell the actual capabilities of large language models and generative AI in their current state. Nonetheless, the surrounding narrative has become so dominant that it divides opinions into near-religious camps. On one side lie the techno-optimists, who are convinced that AI will transform every industry and replace most jobs before the decade ends. On the other side are the skeptics, who remain wary of the flood of unoriginal AI-generated content, misinformation, and bots that saturate online spaces and distort political discourse. Both sides might sometimes exaggerate, but AI’s diffusion and influence are undeniable. In practice, AI algorithms now mediate everything from what we buy and who we date to the jobs, loans, and healthcare we access, often perpetuating biases and causing harm in ways we barely understand.

European governments and policymakers find themselves in the middle. They must act, yet it is often unclear what exactly to act on. The EU’s Draghi Report warns of Europe’s vulnerability, urging massive investment to stay competitive in the AI race.[v] With the EU’s AI Continent Action Plan, aiming to position Europe as a global leader in AI, the path has been laid. Policies and initiatives have been implemented to address growing concerns regarding Europe’s competitiveness: more computing power, better access to high-quality data, a stronger AI talent base, increased adoption, regulatory simplification, and, to top it all off, a €200 billion investment pool. Time will tell which of these pillars will prove most effective.

At the same time, the governments must also confront the darker consequences of the same technology, such as deepfakes, discrimination, job displacement, and the potential for mass surveillance through facial and emotion recognition. How do we ensure that we safeguard citizens from all this while not stifling businesses’ capacity to innovate and create AI solutions that genuinely advance fields such as transportation, energy, healthcare, and agriculture? The phrase “harnessing the benefits while mitigating the risks,” despite being sensible, has been repeated almost ad nauseam since the first AI Safety Summit at Bletchley Park in 2023. Nevertheless, this remains the overarching aim and main challenge of global AI governance.

This balance has become especially challenging as the US, a central player in global AI research and deployment, has increasingly positioned artificial intelligence as a strategic national priority with implications for economic and geopolitical power—while discussions about AI safety are largely not on the table, at least not on ones the US appears willing to sit at. However, we have seen significant progress in formalizing AI governance. The OECD AI Principles (2019) established the first intergovernmental framework for responsible AI, the UNESCO Recommendation on AI Ethics (2021) articulated shared global values, and the G7 Hiroshima Process and AI Safety Summits (2023–2024) brought heads of state together to align their approaches to AI safety and oversight. The Safety Summits, later rebranded as the “Action Summit” in Paris 2025 and the recent “Impact Summit” in India, are emblematic of the current state of AI governance, in which the scales have tipped toward innovation and competitiveness, leaving long-term safety and equity concerns increasingly peripheral.

Notably, the UN’s Global Digital Compact, adopted at the 2024 Summit of the Future in New York, was annexed to the Pact for the Future and commits all 193 UN Member States to cooperate in the digital domain, including AI governance, human rights online, data governance, and digital inclusion. All of these multilateral efforts, while fragmented and largely nonbinding, have gradually paved the way for more concrete regulation.

This brings us to the EU AI Act—the first comprehensive, legally enforceable framework for AI. The Act classifies AI systems by risk level, from minimal to high and unacceptable, and assigns obligations accordingly. Will this first-of-its-kind regulation hurt innovation? The concern remains among some European governments and organizations and continues to fuel the so-called “stop-the-clock” movement calling for a pause or simplification of the Act’s implementation. It is still very early to predict the true effectiveness of the EU AI Act. Its success will depend on its interpretation, enforcement, and ability to evolve alongside the technology it governs. The AI Act may not be perfect, but in a time of fragmented global governance and growing technological asymmetry, it at least represents something tangible: an attempt to translate years of normative debate and principles into law and to anchor AI deployment within a framework of democratic accountability.

Europe can, without a doubt, lead the race on governance. Where other regions rely on voluntary principles or corporate self-regulation, the EU has moved ahead with concrete legislation and long-term strategic planning. Yet governance alone does not build capacity. To remain relevant and reduce dependence on imported tech stacks, Europe is also investing heavily in AI infrastructure. Under the banner of digital sovereignty, this has become a defining theme of European tech policy.

Still, it is fair to ask whether this focus on scale and “hardware competitiveness” is justified or even necessary. Are we overestimating the number of European actors that need access to such vast computing resources? If Europe maintains a focus on open science and open-source approaches, many AI tools will remain accessible for adaptation and training with local data. But making meaningful use of them requires skilled talent and resilient research institutions, both of which continue to receive only a fraction of the funding directed toward large infrastructure projects.

For Europe to sustain an independent and responsible AI ecosystem, economic incentives and regulatory foresight must align. The goal should not necessarily be to compete on computing capacity or market share but to foster innovation that reflects the values of transparency, accountability, and human rights. This is especially true because the former is not predictive of a thriving society.

Ultimately, Europe’s path should, and will likely be, less about winning the AI race and more about defining how it can coexist with democratic institutions, workers, and citizens. The task is not to match the speed of others because haste makes waste, even if that waste only becomes apparent down the line. While it remains unclear what scale of catastrophe it will take for some to realize this; Europe at least seems more intent on learning the lesson before it is forced upon them.

[i] Ashish Vaswani et al., “Attention Is All You Need,” Advances in Neural Information Processing Systems 30 (2017).

[ii] Jared Kaplan et al., “Scaling Laws for Neural Language Models,” January 23, 2020, arXiv:2001.08361 [cs.LG], https://doi.org/10.48550/arXiv.2001.08361.

[iii] See International Monetary Fund, Global Financial Stability Report: Shifting Ground beneath the Calm (International Monetary Fund, October 2025), chap. 1, fig. 1.1, https://www.imf.org/en/publications/gfsr/issues/2025/10/14/global-financial-stability-report-october-2025; Hannah Rubinton and Bontu Ankit Patro, “Tracking AI’s Contribution to GDP Growth,” Federal Reserve Bank of St. Louis, January 12, 2026, https://www.stlouisfed.org/on-the-economy/2026/jan/tracking-ai-contribution-gdp-growth; Felix Richter, “Wider Tech Sector Led S&P 500 To Another Double-Digit Gain,” January 6, 2026, Statista, https://www.statista.com/chart/30318/sector-contributions-to-sp500-return/?srsltid=AfmBOopxONmja1Hm86JXzljeBYFWyDDUjXhnY244WWpxCPE32QrjvdHS&utm.

[iv] Shana Lynch, “Andrew Ng: Why AI Is the New Electricity,” Insights by Stanford Business, March 11, 2017, https://www.gsb.stanford.edu/insights/andrew-ng-why-ai-new-electricity; Brad Smith, “The Next Great GPT: Advancing Prosperity in the Age of AI”, Microsoft Blogs, October 29, 2024, https://blogs.microsoft.com/on-the-issues/2024/10/29/the-next-great-gpt-advancing-prosperity-in-the-age-of-ai/.

[v] Mario Draghi, The Future of European Competitiveness (European Commission, 2024), https://commission.europa.eu/topics/competitiveness/draghi-report_en.