Europe�s AI dilemma just got sharper. Siemens chief executive Roland Busch says it would be a �disaster� if the EU prioritized building sovereign AI infrastructure over rapidly deploying today�s best tools to lift growth. His warning lands as Brussels prepares a �tech sovereignty package� to cut reliance on US providers and amid fears of geopolitical decoupling that could leave European industry exposed.
Growth now, sovereignty later?
Busch argues the EU�s approach is �completely miscalibrated� compared with the US, where lighter-touch oversight has created �a fast flowing river� of adoption. Europe, by contrast, resembles �standing water,� he told the Financial Times, cautioning policymakers not to �throttle your innovation speed for the sake of creating sovereignty.�
Siemens is putting �1 billion into industrial AI – from a virtual shop floor manager to automation and digital engineering – but is prioritizing investment in the US and China before Germany. Busch�s central critique: rules fail to distinguish consumer protection from business-to-business use, where companies can contractually govern data sharing. �If you have two companies and they make a contract on how to share data, why would I need to regulate the way companies are sharing that data?� he asked.
Counterpoint: Dependency is already costing Europe
Yet the case for strategic autonomy isn�t hypothetical. European cloud providers say they�re being squeezed after Broadcom�s acquisition of VMware. The CISPE industry association filed a competition complaint in Brussels alleging price hikes of 800�1500%, forced bundling, and the termination of a key VMware partner program in Europe. Dutch and Danish cloud groups report �exponential� cost increases and an �extremely critical high-pressure situation,� warning this could push customers to US hyperscalers and undermine sovereignty goals.
That real-time dependency shock strengthens the argument that Europe needs leverage – through standards, portability, and viable homegrown options – even as it accelerates deployment.
Balancing speed with leverage: A European middle path
Some European founders are proposing instruments to square the circle. Mistral AI�s CEO has floated a revenue-based content levy on commercial AI providers operating in Europe – domestic and foreign alike – to fund creators and provide legal certainty for training data. The company says it is investing �4 billion in European infrastructure to train models on EU soil, positioning the levy as a pro-innovation, pro-culture compromise.
Critics question whether a levy could raise costs for startups and users, but the idea underscores a broader point: Europe can design frameworks that encourage deployment while reclaiming bargaining power over data and compute.
Mind the capital stack: Who captures the upside?
Even if Europe speeds up adoption, who benefits? BlackRock CEO Larry Fink warns AI�s gains could concentrate among firms with data, infrastructure, and capital. In his shareholder letter, he argues the �companies with the data, infrastructure and capital to deploy AI at scale are positioned to benefit disproportionately,� risking a repeat of past wealth concentration dynamics unless more people and smaller firms can participate.
For Europe�s industrial mid-caps, that�s a nudge to move fast – but also to secure favorable terms and portability when partnering with large vendors. Otherwise, they could end up as price-takers in critical infrastructure and data.
Regulatory split-screen: US speed vs EU certainty
Across the Atlantic, the policy climate remains comparatively permissive. A recent framework associated with Donald Trump emphasizes narrow AI rules (child safety, content controls) and industry-led standards while rejecting a new federal watchdog – exactly the �lighter� environment Busch says accelerates adoption. By contrast, the EU�s landmark AI regulation, now in force, faces implementation friction and industry pushback.
The strategic trade-off is becoming clearer. The US is optimizing for velocity; Europe is optimizing for governance and resilience. Businesses operating in the EU can�t wait for perfect sovereignty, but nor can they ignore lock-in risks.
What executives should do now
European boards can adopt a dual-track strategy:
- Deploy best-in-class AI immediately for productivity and quality gains, with strong internal data governance.
- Negotiate exit-friendly contracts (interoperability, data portability, clear switching terms) with major vendors.
- Pilot European and open models where performance is sufficient, especially for industrial and domain-specific use.
- Engage in standards and policy debates (e.g., content levies, cloud competition) to shape a level playing field.
Busch is right that waiting for �AI factories� to be built at home could stall growth. But the Broadcom-VMware fallout shows reliance without leverage is a strategic liability. Europe�s best answer may be pragmatic speed today – with portability, bargaining power, and targeted sovereignty embedded by design.