Global AI competition has entered a new phase. While Washington and Brussels build 'Pax Silica' around infrastructure, Beijing is developing a rival coalition around governance. At first glance, this suggests the emergence of a bipolar model. But a closer look, says Elena Yurieva, reveals a far more complex picture
A few years ago, almost no one would have predicted that artificial intelligence would reshape the geopolitical board. The fragile balance of power that took shape after the collapse of the Soviet Union had already been unsettled by Russia's war against Ukraine. AI has given the major players new pieces to move: computers, chips, models and the infrastructure required to sustain them. Now the world is searching for a new equilibrium.
Over recent months, several developments have reinforced the perception that the emerging AI order is becoming bipolar. In December 2025, the US State Department launched Pax Silica, a coalition of allied nations focused on securing the physical foundations of AI: semiconductors, critical minerals, energy and computing infrastructure. By its second summit in June 2026, the initiative had grown to 24 signatories. US private AI investment reached $285.9 billion in 2025, more than 23 times the $12.4 billion invested in China, though private investment figures alone likely understates China's total AI spending, given its government guidance funds.
Pax Silica is a US-led coalition of allied nations focused on securing the physical foundations of AI: semiconductors, critical minerals, energy and computing infrastructure
In June 2026, after prolonged deliberation and the resolution of internal disagreements with France, the European Union joined Pax Silica. On 17 July, in Shanghai, China announced the World AI Cooperation Organisation, an intergovernmental body with 29 founding members built around governance, international cooperation and institutional coordination, described by Beijing as the world's first intergovernmental organisation devoted specifically to AI.
The European Union spent most of this period outside both coalitions. It observed Pax Silica's December 2025 launch from a distance and declined to sign. That changed on 23–25 June 2026, when EU ambassadors authorised the European Commission to sign the Pax Silica Declaration on the Union's behalf.
US Secretary of State Marco Rubio had made Washington's argument explicit in Senate testimony days earlier, describing Pax Silica as a 'global consortium' covering 'all of the elements the AI supply needs' and stressing that US leadership 'is not permanent'. The European Commission's own announcement framed EU membership as compatible with, even supportive of, the technological sovereignty package it had adopted only weeks earlier, including Chips Act 2.0. Whether that framing survives contact with the reality of coalition membership remains an open question.
Europe has continued shaping the regulatory environment in which AI develops. The EU AI Act remains the world's first comprehensive binding legal framework for artificial intelligence. At the same time, regulation alone cannot guarantee geopolitical influence. Since April 2025, the AI Continent Action Plan has mobilised €200 billion through its InvestAI initiative, with cumulative datacentre investment projected at €176 billion by 2031. The EU Council had already adopted its Declaration for European Digital Sovereignty in November 2025, a month before Pax Silica launched.
Much of the physical infrastructure now rising on European soil remains under non-European ownership, including by tech giants Microsoft and Nscale
Whether these investments will ultimately produce genuine technological sovereignty remains uncertain. Much of the physical infrastructure now rising on European soil remains under non-European ownership. Microsoft and Nscale are building a gigafactory-scale AI campus in Sines, Portugal. Analysts conclude that Europe's first gigafactory-scale facilities are unlikely to be EU-funded, EU-owned, or built around European frontier-model developers.
The pattern extends beyond infrastructure to capital itself: EU-based private equity and venture capital firms invested $6.8 billion in local AI companies in 2025 but sent nearly as much – $6.21 billion – into AI companies in the United States. Large American investment houses have already formed partnerships with OpenAI and Anthropic; European firms, with few exceptions, have not.
At first glance, every side's position looks declared and clear. The United States seeks to secure the technological foundations of AI. China is willing to share the institutions through which AI will be governed. Europe is attempting something more unusual: participating in one coalition while preserving enough strategic autonomy to continue building another centre of influence. None of the three is playing the same game, and that may soon lead to one of the following.
The EU will most likely remain formally inside Pax Silica without pausing its own AI regulatory work. It will keep investing in building its own AI sovereignty, infrastructure and technological base. What remains unknown is how Pax Silica's other partners will come to view that dual position over time.
Much depends on how the rest of the world reads China's move; whether it drives further expansion of Pax Silica and the World Artificial Intelligence Cooperation Organization, and who ends up joining which coalition, perhaps both at once.
We cannot rule out the possibility that some countries will choose to build their own regional alternative, based on governance, technology, or both.
Which of the above proves right remains uncertain, as well as which of the possible coalitions will become more influential. What is certain is that the world is engaged in a long geopolitical game with formidable opponents.
The board has been reset. The United States has secured its coalition. Europe has revealed its position. China has built its own institutional alternative. The impending moves won't define who leads the AI era, but whether the world still agrees on the rules of the game.
The impending moves by the US, Europe and China won't define who leads the AI era, but whether the world still agrees on the rules of the game
One conclusion already seems difficult to avoid. Instead of converging toward a common institutional framework, the international system is moving in the opposite direction. Infrastructure is being organised through one coalition, governance through another, while technological sovereignty increasingly becomes a national project.
A single international institution capable of coordinating technological development and governance, built on shared principles and mutually accepted limits, would not eliminate strategic competition. States would continue competing economically, scientifically and geopolitically. But competition inside a shared framework is not the same as competition between rival ones.
