By Publius (of the 21st Century) — No. 7

The global race for artificial intelligence has matured far beyond a contest of corporate engineering; it is now a defining geopolitical arena of the twenty-first century. As the United States and China contend for technological supremacy, the European continent—comprising both the institutional weight of the European Union and the post-Brexit United Kingdom—finds itself caught in a structural pincer. To navigate this bipolar transition without forfeiting its autonomy, Europe must weigh two distinct temptations: the Vassal Lure of total alignment with Silicon Valley, and the Siren's Call of Beijing's state-centric alternative.

The Vassal Lure: Alignment with Silicon Valley

The United States offers Europe a comfortable, value-aligned partnership framed in the language of democratic solidarity and NATO interoperability. This is the Vassal Lure. Washington's approach—a private-sector-led, venture-backed ecosystem protected by the “small yard, high fence” export-control doctrine—seeks a unified Western front against China's technological ascent.

For Europe, uncritical alignment with this model carries a long-term cost: dependency without leverage. Because foundational AI infrastructure, hyper-scale computing, and frontier models remain concentrated overwhelmingly in American hands, transatlantic alignment without countervailing investment risks locking European enterprises into permanent reliance on U.S. computing monopolies. Washington's framework invites Europe to the table, but chiefly as consumer and security dependent—constraining rather than enabling Europe's own industrial and technological ambitions.

The Siren's Call: Beijing's Alternative

China has offered a competing vision. President Xi Jinping's Global AI Governance Initiative (announced in October 2023, since supplemented by a Global AI Governance Action Plan and proposals for a World AI Cooperation Organization) explicitly calls for AI development to be open, inclusive, and non-discriminatory, and insists AI must not become “a tool for maintaining hegemony or seeking unfair advantages.” Beijing has paired this rhetoric with a related but distinct initiative—the Global Civilization Initiative—that invokes civilizational pluralism more broadly; the two are sometimes conflated in Western commentary, and the analytic distinction matters for anyone reading Chinese diplomatic language closely.

By advocating a UN-centered model of AI governance, in which “the United Nations play[s] the primary role,” China offers Europe an alternative to Washington's exclusive coalitions. It appeals to Europe's traditional commitment to multilateralism and to the EU's own “strategic autonomy” instinct. Beijing's implicit pitch: why bind Europe's economic future to American monopolies when a state-backed but nominally sovereignty-respecting ecosystem is on offer?Beneath the inclusive rhetoric, however, lies a different reality. China's domestic model of “secure and controllable” AI rests on state censorship and algorithmic alignment with authoritarian priorities—a governance philosophy fundamentally at odds with what Europe claims to defend. Beijing's diplomatic playbook has also relied at times on asymmetric economic pressure exploiting policy fractures between individual EU member states.

Embracing the Siren's Call too closely risks fragmenting the Western alliance and leaving Europe exposed to targeted economic coercion.

Structural Realignment: A More Qualified Convergence Than It Appears

A dual-track strategy—engaging both Washington and Beijing without fully aligning with either—is the logical hedge for Europe. Executing it, however, requires an internal cohesion that European politics has historically struggled to sustain, and recent developments suggest that cohesion is more contested than a straightforward reading of “the Brussels Effect” implies.

The post-Brexit UK cannot easily diverge from the EU's regulatory gravity: because firms build to the highest applicable standard to preserve market access, the EU AI Act functions as a de facto continental baseline regardless of UK domestic preference. But two qualifications are worth noting. First, the EU AI Act itself is not a fixed, escalating edifice—by mid-2026 the Commission's “Digital Omnibus” process had extended compliance deadlines for high-risk AI systems, reflecting real friction between regulatory ambition and industry and member-state pushback. Second, the UK's own posture shifted materially in February 2025, when its AI Safety Institute was renamed the AI Security Institute and its remit narrowed from broad AI ethics (bias, free expression, transparency) toward security-coded threats—cyberattack, fraud, and criminal misuse. This is not simply UK pragmatism bridging Brussels bureaucracy; it is a substantive re-prioritization that puts London closer to Washington's security framing on some dimensions even as market-access logic pulls it toward Brussels on others. The “unspoken alliance” between London and Brussels is real but narrower and more conditional than a single regulatory convergence narrative suggests.

Conclusion: Holding the Middle Ground

To avoid being pulled toward either the Siren's Call or the Vassal Lure, Europe cannot remain a passive consumer or a pure bureaucrat. For lack of its own capability and capacity, Europe cannot continue to rely on regulating technologies that it neither invented nor fully understands—a posture that mistakes rule-writing for power. Hedging works only if Europe holds genuine leverage.

The EU and UK need their shared regulatory weight to function as a tool of market access, not merely compliance overhead—one that obliges both American and Chinese firms to adapt to European terms. Simultaneously, Europe needs to build sovereign computing capacity, cloud infrastructure, and its own frontier-adjacent models, rather than treating regulation as a substitute for capability. In the AI hedge, playing both cards is the only viable path—but staying at the table requires Europe to hold chips of its own, not merely a rulebook.

Addendum: Capability, Capacity, and Strategic Freedom

The question is not whether Europe should sever itself from American technology or seek refuge in Chinese alternatives. Either course, pursued as exclusive dependence, would merely exchange one vulnerability for another.

Former Australian Prime Minister Kevin Rudd’s account of Xi Jinping clarifies why the Chinese alternative must be considered strategically, not merely commercially. Under Xi, technological and industrial leadership forms part of a long-term project to reshape international power, rules, and dependencies in China’s favor. But dependence on American technology is not politically neutral either. Frontier models, advanced chips, cloud infrastructure, and even the continuing availability of services remain subject to decisions made outside Europe.

For the European Union and the United Kingdom alike, genuine autonomy therefore requires more than technological competence. It rests on four closely connected foundations.

The first is AI capability: the scientific, engineering, entrepreneurial, and institutional ability to design, train, adapt, audit, secure, and govern advanced systems.

The second is AI capacity: sufficient computing power, energy, cloud infrastructure, capital, data, and deployment scale to make those capabilities consequential throughout government, defense, industry, research, and society.

The third is transnational collaboration. No single European country, including Germany, France, or the United Kingdom, can efficiently reproduce the full research, computing, industrial, and commercial ecosystem now concentrated in the United States and China. European sovereignty will therefore depend on whether national laboratories, universities, companies, governments, and capital markets can cooperate across borders without allowing national prestige, incompatible procurement rules, or industrial rivalries to paralyze common projects.

The fourth is financing, which is inseparable from collaboration. Europe does not merely need more research grants. It needs patient, large-scale, risk-tolerant financing capable of carrying promising systems from the laboratory through model training, infrastructure construction, commercialization, and continent-wide deployment. Fragmented national subsidies will not create European scale. Joint financing must support joint institutions, shared infrastructure, and commercially viable enterprises capable of competing beyond their domestic markets.

The troubled Franco-German experience with the Future Combat Air System should serve as a warning. If disputes over leadership, intellectual property, national work shares, and corporate control are reproduced in artificial intelligence, European AI sovereignty will be forfeited before it is established. A collection of nationally protected projects, each too small to compete globally and too politically constrained to integrate, would amount to technological fragmentation disguised as sovereignty.

Capability without capacity produces excellent laboratories but continued dependence. Capacity without capability produces rented power. Collaboration without workable governance produces delay and paralysis. Financing without shared strategic purpose produces scattered subsidies rather than durable strength.

The objective should not be autarky, nor a futile attempt to duplicate every American and Chinese platform. It should be credible strategic optionality: European systems strong enough to guarantee continuity in essential domains, combined with diversified access to American, Chinese, and open-source technologies wherever cooperation remains advantageous.

Europe must be able to cooperate without being compelled, purchase without becoming captive, and refuse without becoming technologically disabled. The siren’s call and the vassal’s lure are dangerous for the same reason: both promise immediate power at the price of future freedom.