China is pitching its AI models to Europe, but is it enough to avoid 2-horse race with US?

As the US and China solidify their leadership in artificial intelligence, Europe finds itself in a precarious position. The race to dominate the AI landscape is becoming a battle over international order. In June, the United States expanded its Pax Silica framework, adding 10 more partners to secure supply chains for critical technologies like microchips and AI, bringing the total to 24.

Meanwhile, 29 countries signed an agreement in Shanghai to establish the World AI Cooperation Organisation, a Chinese-led initiative aimed at the Global South.

According to the 2026 Global AI Innovation Index, published by China's Institute of Scientific and Technical Information and Peking University, the US and China still lead far ahead of other countries. The US scored 78.44 points, while China came in second with 60.49. Britain was placed third with 39.99 points, significantly lagging behind the other two nations. Even Germany and France, ranked sixth and seventh respectively, were over 20 points behind China.

Beijing is actively promoting its open-source AI models to European capitals as a more affordable alternative to US technology. Last month, China's ambassador to Britain, Zheng Zeguang, called for cooperation on this issue in an article for The Guardian, highlighting models like DeepSeek as "global public goods" that lower barriers to innovation.

In June, China's ambassador to the European Union, Cai Run, made a similar appeal at the Europe Forum in Brussels, stating that China's open-source AI models, combined with Europe's strengths in basic research and regulation, could be a powerful combination.

From Beijing's perspective, engaging with European partners is crucial to avoid a complete US-China confrontation, as Cui Hongjian, a research fellow at Beijing Foreign Studies University's Country and Area Studies Academy, explained. "A so-called G2 structure seems to be forming in the AI field," Cui said, referring to the US and China's stronger systemic capabilities.

However, Cui also acknowledged the possibility of a "G3" structure, stating that while AI talks between China and the US are ongoing, Beijing also desires a similar arrangement with Europe. "Europe acknowledges this disparity, but long-term trends leave open the possibility of a G3," Cui added, advocating for a balanced trilateral structure instead of Europe becoming an American digital colony.

To become a genuine "G3" player, Europe must build its own systemic capabilities, such as basic computing power and infrastructure. However, achieving this in the short term seems challenging, Cui warned. Europe could preserve its technological sovereignty by adopting a model compatible with China's open-source infrastructure. Some European companies have already started adopting Chinese AI models to reduce development costs and lessen reliance on US technology. Siemens, for instance, reportedly uses models like DeepSeek and Z.ai to achieve this flexibility.

Despite the potential benefits of open-source models, they may not be sufficient to bridge Europe's significant infrastructure gap. Sebastian Contin Trillo-Figueroa, a geopolitics strategist specializing in EU-Asia relations, described the current situation as an "orbital bipolarity," where two centers of gravity draw all other actors into their orbit.

While China's open-source push cannot guarantee technological sovereignty for European countries, it could serve as a valuable bargaining chip against the US. Trillo-Figueroa emphasized that Europe's real leverage lies in high-value industrial assets and rich proprietary data sets in various sectors, including healthcare, manufacturing, energy, and aerospace.

Geopolitics strategist James Downes suggested that Europe's heavy reliance on US cloud infrastructure has limited its options. Beijing's open-source pitch might create tactical space for selective cooperation, but Downes cautioned that without faster investment in computing power, capital, and scale, Europe and the UK risk becoming mere passengers in the broader AI race.