Deep pockets, DeepSeek, and the EU’s digital drift
The US and China are racing to dominate artificial intelligence, while Europe remains a cautious spectator. If the bloc wants a future in tech, it must go big on industrial AI—or go home
The news that a Chinese start-up, DeepSeek, launched a low-cost AI model in late January has sent shockwaves through technology markets, which fear this might threaten America’s AI dominance. This came less than a week after President Donald Trump announced on January 21st that four tech firms would invest a spectacular $500bn in American AI infrastructure, the Stargate project. European firms, on the other hand, are absent from this technology race. To position itself as a serious competitor, the European Union should invest in sustainable AI across various sectors, including healthcare, education, climate and energy, mobility, and urban planning.
The EU’s digital landscape
Although the EU has few top-tier technology companies, it has robust, digitised, technology-driven industries that provide goods and services on a global scale. It is home to 6 of the top 20 carmakers in the world by market capitalisation, one of the key pillars for mobility and smart cities. It also has top companies in the electricity, defence, aerospace, insurance and pharmaceutical sectors. It has the second largest market in the world in nominal terms, behind the United States.
Beyond the size of its market, the EU’s strength also lies in promoting social rights and environmental protection across these key global sectors. These are a core component, for example, of the EU’s 2021 Global Gateway strategy to invest in smart, clean and secure infrastructure globally. Companies ranking high on environmental, social and governance principles perform just as well as others, if not better, in the medium to long term.
However, EU competitiveness is falling behind as some of its business strengths—market power and sustainability—are eroding. Its advanced sectors began their lead during the pre-digital era. Now they lag America and China in innovation and maintain a conservative mentality.
They are also facing increasing competition from US big tech companies expanding their portfolio to areas where European firms used to have the upper hand, such as finance (today challenged by Apple on payments), telecommunications (Meta’s WhatsApp active users represent one-third of the world population), energy (Microsoft and Amazon announced they are investing in their own nuclear energy provision) and healthcare (Amazon even has its own clinics).
Lastly, European firms also bear a heavier regulatory burden than the US, a point that American tech firms often contest. Just as important is the uneven implementation of these rules among EU member-states, which results in a fragmented environment that prevents start-ups from scaling.
AI and European competitiveness
AI can help these industries acquire more market power by enhancing scalability. These industries also generate valuable data, which, along with skills and computing power, is essential to help train AI systems to become better. Consulting firm McKinsey projects that generative AI can add globally €680bn per year in advanced manufacturing (automotive, aerospace and defence), energy, pharmaceuticals and insurance by improving customer operations, marketing, sales, software engineering, and research and development (R&D). A quarter of the top 20 companies in these sectors are from the EU.
The energy sector in particular could be transformed by AI. EU policies (from incentives to regulation and fines) have already encouraged member states to save energy and accelerate the transition to green energies, consolidating the EU’s advantage against the US and chasing the global leader China with more solid business models. But this can become even more of an asset for the EU. AI can significantly improve energy efficiency through predictive analysis, optimising real-time energy consumption and adjusting demand based on environmental and operational data. This is already applied to smart grids and energy management systems. For example, Google slashed up to 40% of the energy it used to cool its data centres by applying machine learning. AI innovations can therefore reduce global greenhouse gas emissions by 5% to 10%.
AI also facilitates the integration of renewable energy sources into energy grids. It can predict solar or wind power fluctuations, so it can help balance the energy supply more effectively across the grid.
In pharmaceutical and medical products, the potential is also remarkable. The EU is the second-largest pharmaceutical market in the world after the US. To maintain this lead, it can harness generative AI to improve the speed and the quality of R&D for drug discovery. For instance, AI technologies accelerated the development of covid-19 vaccines, reducing the time from conception to distribution from years into months.
Recommendations for the EU
In 2024, the EU published two new reports on its economy: one on the future of the single market by the former Italian prime minister, Enrico Letta, and one on European competitiveness by the former European Central Bank president, Mario Draghi. The reports advise the EU to embrace technology and innovation under a new industrial strategy, increase R&D funding, and modernise finance. Financial market reform should include eliminating (or reducing) capital market segmentation, reviving the securitisation market to boost the banking sector's financing capacity and moving towards a single common regulator. This would help overcome Europe’s main obstacles to realising its digital potential: regulatory fragmentation, the absence of a single digital market, and the lack of a developed venture capital industry, which limits financing.
Currently, AI is primarily developed by big tech firms, which European users then buy and adapt. Start-ups need these firms’ AI models, computing power and data to grow. Conversely, big tech can gain innovative ideas and projects from start-ups. Both can benefit from collaboration, so the EU should encourage such projects.
The EU should also favour a pro-innovation implementation of its laws. For example, it could establish “regulatory sandboxes” to provide safe spaces for companies to experiment with AI-powered services before they are widely deployed, giving both companies and regulators invaluable insights into their real-world impacts.
Lastly, it should invest in specialised smaller and midsize generative AI models. Model size is the number of parameters a machine learning system uses to train. Smaller models, designed for certain tasks, are trained on less data, need less computing power and even avoid the need to use the latest chip or equipment. This means they can used more efficiently in specific industries, such as energy, advanced manufacturing and pharmaceuticals.
The EU should prioritise developing smaller AI-powered models specialised in key sectors while also tackling high regulatory uncertainty, compliance costs and fragmentation
In other words, the EU should not focus on catching up on AI hardware and infrastructure, as it is doing currently. Instead, it should prioritise developing smaller AI-powered models specialised in key sectors while also tackling high regulatory uncertainty, compliance costs and fragmentation.
Europeans should take advantage of smaller industry-specific AI models to circumvent the bloc’s lag in developing digital technologies, given its infrastructure gaps. It could then emerge as a leading global competitor in key AI-powered sectors such as health, energy or mobility, showcasing a sustainable technological and economic revolution that rivals the US and China.
The European Council on Foreign Relations does not take collective positions. ECFR publications only represent the views of their individual authors.