Washington and AI: A Growing Pressure Lever on Europe

AI, an Essential Pillar for Europe under American Influence

Artificial intelligence (AI) has become an indispensable component for European companies, playing a crucial role in their productivity and competitiveness. However, access to this technology relies on a heavily concentrated value chain that is largely under the control of the United States. In an increasingly tense geopolitical climate, a political decision from Washington could be enough to disrupt the prices and availability of AI for European businesses.

A Fictional Scenario That Could Become Reality

On February 25, 2027, OpenAI completed a historic funding round, raising over $100 billion from giants such as Amazon, Nvidia, SoftBank, and Microsoft. This funding propelled the company's valuation to nearly $850 billion, placing it among the best-funded private companies in the history of technology. Faced with this financial pressure, the challenge for OpenAI is now to transform this race for billions into sustainable profitability.

Simultaneously, Washington is seeking to increase its pressure on European economic players, who are now largely dependent on AI for their efficiency. Cutting direct access to AI would be too costly for American giants. However, multiplying the cost of access to AI by two, five, or even ten could prove to be an effective strategy. In this fictional scenario, such a policy would benefit OpenAI and certain American decision-makers, but would be significantly less favorable to European companies.

The Real Cost of AI, a Well-Kept Secret

Whether from an environmental or economic perspective, there is a total disconnect between the investments made, the value created, and the revenue generated. In 2025, over $400 billion was invested in AI infrastructure, an amount greater than the total cost of the U.S. highway network built over 35 years. Cumulative spending on infrastructure and computing capacity for AI could exceed a trillion dollars by 2030, even as returns on investment remain uncertain.

Currently, prices are artificially kept low to maximize the rapid adoption of AI and create dependency. In other words, the physical resources committed are increasing much faster than the value actually charged. The initial doses are almost free...

Why Washington Could Influence AI Prices in Europe

Regarding intent, the United States is increasingly using critical technologies as instruments of economic and strategic policy, whether it concerns semiconductors, export controls, or extraterritorial regulations. In a context of heightened trade tensions, industrial rivalries, and pressures on supply chains, access to computing could become an additional lever of influence. Rather than abruptly cutting access to AI, a targeted increase in costs for heavily dependent companies would constitute a more subtle, economically viable, and politically defensible tool.

As for the possibility, it stems directly from the very structure of the value chain. From chip design to cloud hosting, including AI models, the market is dominated by the United States. Nvidia controls about 90% of the GPUs dedicated to AI training, while AWS, Microsoft Azure, and Google Cloud concentrate most of the global computing capabilities. In other words, with the notable exception of Dutch lithography with ASML, which holds 100% of the market for advanced EUV machines, and a significant portion of Taiwanese processor manufacturing with TSMC, which holds 90% of the market for chips <7nm, the critical AI value chain is largely anchored in the American tech ecosystem. This concentration creates a structural lock-in: the more dependency sets in, the harder it becomes to change suppliers or jurisdictions.

For example, while waiting for Mistral's European data centers to be fully built and operational, the company currently relies on existing clouds, notably Microsoft Azure and Google Cloud, to host and run its models.

An Unavoidable Strategic Question for Leaders

Beyond the hypothesis of future Washington control over AI pricing in Europe, the essential point for companies lies elsewhere: AI must be recognized as critical infrastructure, access to which depends on a concentrated, physically constrained, and legally situated value chain.

In a context of international turbulence, techno-geopolitics has become a strategic parameter in its own right that leaders can no longer afford to overlook.