Seyithan Ahmet AtesThe global race for artificial intelligence (AI) is increasingly taking shape as a two-bloc order: the United States and China on one side, and Europe, the Central Powers and developing countries on the other.
German Chancellor Friedrich Merz demanded that Europe should not cede the leading role to China and the USA, while French President Emmanuel Macron advocated giving preference to European companies in public digital projects.
A growing body of evidence indicates that China and the United States are building structural advantages across the entire AI value chain, spanning talent, computing power, capital, cutting-edge models, cloud infrastructure, and industrial ecosystems.
Together, they employ roughly 70 percent of the world's leading machine-learning researchers and account for about 90 percent of global computing power. They also attract the vast majority of AI investment — more than twice as much as all other countries combined.
In previous technological revolutions, countries outside the leading group gradually adopted new technologies and caught up. With AI, this path may be blocked.
Progress in AI depends not only on brilliant engineers or excellent universities, but also on chips, data centres, cloud platforms, affordable electricity, and substantial capital investment.
Countries lacking these foundations could not only fall behind; they could become dependent on systems developed, hosted, and controlled elsewhere.
The US remains the strongest player in several crucial areas. It hosts 5,427 data centres, more than ten times as many as any other country, and continues to lead the world in private AI investment.
Private AI investment in the US is estimated to be 23 times higher than in China; in generative AI, American investment significantly exceeds the combined total of China and Europe.
However, China has built a strong alternative base. The country leads in the number of scientific publications, citations, and granted patents. The United States continues to produce more high-impact patents and cutting-edge models, but China's scientific and industrial scaling is rapidly narrowing the gap.
In 2025, the US produced 59 notable AI models, compared with China's 35. At the same time, China's share of the 100 most-cited AI papers rose from 33 percent in 2021 to 41 percent in 2024.
Too much rhetoric, too little implementation
Europe is not absent from the AI race, but it is weaker than its political rhetoric suggests. The European Union has positioned itself as a regulatory power, most visibly through the EU AI Act.
Across Europe, including the United Kingdom, there are strong researchers, excellent universities, and major companies such as Helsing and Mistral. Nevertheless, Europe has limited success in translating research into globally competitive businesses.
European startups often relocate to the United States to access capital, customers, and computing power. Many leading European AI researchers pursue graduate studies in the US and then remain there.
Even more serious is the infrastructure gap. Mistral CEO Arthur Mensch has warned that Europe's reliance on American cloud infrastructure for training AI models constitutes a strategic vulnerability.
Mistral is considered the only European company whose language model, according to experts, can compete with ChatGPT and other US systems, making it a crucial building block of Europe's claim to technological independence.
Even if European companies develop strong models, the underlying training infrastructure, chips, and cloud capacity often remain under the control of non-European companies.
The GAIA-X project exposes the limitations of the current European approach. Launched in 2019 as a Franco-German cloud initiative, it was presented as a major step towards European digital sovereignty.
In practice, however, it has failed to deliver the promised scale of operational infrastructure. European cloud providers hold only a small share of the continental market, and this share is shrinking rather than growing.
Closing this gap would require sustained, coordinated investment on a scale that is not currently politically feasible at the EU level.
Asia's hardware advantage, Europe's structural weakness
Outside the US and China, several Asian economies have identified clear entry points into the AI economy, with South Korea and Taiwan the most prominent examples.
The boom in AI chips has propelled Taiwan and South Korea ahead of the United Kingdom in global market rankings.
This development is driven by companies at the heart of AI infrastructure: Taiwan Semiconductor Manufacturing Company (TSMC), the world's largest contract chip manufacturer, and South Korea's Samsung Electronics and SK Hynix, which dominate key memory markets.
TSMC is now one of the world's largest companies, with a market capitalisation of approximately US$1.8 trillion. Samsung and SK Hynix together have a combined market capitalisation of around US$1.5 trillion.
For comparison, the combined market value of all technology stocks in the Stoxx Europe 600 Index is about US$1.4 trillion.
Asia is benefiting from the hardware foundations of the AI revolution, while Europe remains more focused on regulation, financial services, and fragmented digital initiatives.
This is where middle powers become important. No single middle power can realistically replicate the full AI stack of the United States or China. However, they can combine complementary strengths, such as energy, geography, hardware, talent, industrial applications, and market access.
Electricity prices in Türkiye are roughly a quarter of the European average. As electricity is one of the most significant cost factors for data centres, Türkiye is a leading potential location for data centre investments.
The AI gap is not just an economic problem; it's also a security issue. If the US or China were to deny a country access to AI systems hosted on their territory today, the immediate consequences might still be limited.
Most hospitals, military systems, power grids, and public services are not yet fully dependent on cutting-edge AI.
But that will change. As AI becomes more deeply embedded across critical infrastructure, defence systems, logistics, finance, healthcare, and public administration, reliance on foreign AI platforms will become a strategic vulnerability.
Both the United States and China have already demonstrated, in other areas, their willingness to use technological and economic dependencies as leverage. There is no reason to believe AI will be any different.
Therefore, data centers, cloud infrastructure, and energy policy have become matters of national strategy. Countries that do not control a significant share of the AI infrastructure could find their political room for manoeuvre restricted in the future by those that do.
The concentration of AI power in the United States and China poses a long-term strategic challenge for Europe and other middle powers.
Asia, particularly South Korea and Taiwan, has capitalised on opportunities through its hardware and semiconductor ecosystems. Europe, by contrast, risks falling behind because of high energy costs, weak commercialisation, fragmented cloud infrastructure, and over-reliance on regulation.
The gap between the US-China AI axis and the rest of the world is widening.
If middle powers fail to coordinate their strategies across energy, computing power, hardware, talent, and capital, the AI divide will become a permanent structure of technological dependency.
