Europe doubles down on digital sovereignty: chips act 2.0 and cloud rules spark debate over innovation, independence, and energy costs

The European Commission’s Chips Act 2.0 marks a strategic pivot from primarily supply-side interventions to robust demand-side measures aimed at stimulating domestic uptake of EU-made semiconductors. By incentivizing government procurement—especially from startups—and deploying instruments like public innovation procurement and 'Demand Accelerators,' the EU seeks to create a more balanced semiconductor ecosystem that reduces reliance on US and Asian suppliers. This approach is reinforced by new intervention powers allowing prioritization of critical chip production during crises and rapid fab approvals, signaling a comprehensive effort to strengthen Europe's chip sovereignty through both market stimulation and regulatory oversight.

Europe’s existing strengths in chip manufacturing machinery, research institutions, and open-source development form an underleveraged foundation that the Commission aims to harness by connecting chipmakers more closely with growth sectors such as data centers, cloud providers, and AI Gigafactories. The plan to at least triple EU data center capacity within 5 to 7 years complements this by fostering a sovereign digital infrastructure that naturally drives demand for domestically produced chips. Moreover, the Cloud Sovereignty Framework’s detailed scoring system (SEAL-0 to SEAL-4) prioritizes European control in critical public tenders, favoring providers like OVHcloud and creating opportunities for smaller European players to compete beyond scale alone.

Recognizing that Europe’s semiconductor demand is currently dominated by mature 28/22 nanometer technologies serving automotive and industrial sectors, policymakers advocate shifting support from subsidizing large national champions like Infineon and STMicroelectronics toward nurturing a dynamic fabless AI-chip ecosystem. Proposals include tax breaks, R&D incentives, and scale-up financing to empower startups and design firms, complemented by a pan-European AI Chip Growth Facility co-investing with venture capital in talent-rich but capital-thin regions such as the Netherlands, Poland, and the Baltics. This strategy aims to boost domestic AI chip demand and design capabilities, avoiding the pitfall of subsidizing chip production that ultimately benefits US markets, as cautioned by ASML CEO Christophe Fouquet.

The Cloud and AI Development Act (CADA) ambitiously aims to triple the EU's data center capacity within five to seven years, targeting full sufficiency for European businesses and public administrations by 2035. This expansion is coupled with a strong emphasis on sustainability and innovation, promoting energy-efficient facilities that integrate with Europe's broader green transition and digital twin initiatives. To facilitate this growth, CADA proposes streamlined permitting, improved access to critical resources like land and energy, and the establishment of Experience and Acceleration Centres to foster coordinated AI adoption across Member States.

Central to CADA is the introduction of a unified EU-wide Cloud Sovereignty Framework featuring four assurance levels (SEAL-0 to SEAL-4) that rigorously assess control over technology, supply chains, data processing, infrastructure location, and cybersecurity. This framework empowers Member States and EU entities to conduct tailored sovereignty risk assessments, determining which cloud environments and providers qualify for sensitive workloads. While the framework aims to protect critical applications and sensitive data by excluding providers subject to third-country control—effectively sidelining US hyperscalers due to the US Cloud Act—it balances this with market openness to like-minded partners, avoiding outright bans but demanding deeper scrutiny in public procurement.

The Act's practical rollout is already underway, exemplified by a sovereign cloud contract worth up to €180 million awarded to European providers such as Post Telecom with OVHcloud, CleverCloud, STACKIT, Scaleway, and a Proximus-led consortium. However, challenges remain, including the absence of an operational EU cloud cybersecurity certification scheme—critical for higher assurance levels—leading to reliance on fragmented national certifications and added bureaucracy. Moreover, tripling data center capacity confronts significant energy supply constraints, as noted by Dell'Oro's Alex Cordovil, highlighting that grid limitations, not just regulatory reforms, will be the bottleneck for scaling sustainable infrastructure.

CADA is complemented by the EU's Open Source Strategy, which seeks to scale sovereign open source solutions across cloud, AI, cybersecurity, and semiconductors, thereby reinforcing European ownership and control over digital infrastructure. This strategy supports public administrations through procurement guidelines and investments in skills and startups, fostering a more competitive and resilient digital ecosystem. Meanwhile, the EU is navigating complex debates over sovereignty definitions and AI infrastructure models, with a likely two-tier approach: government-critical AI workloads running exclusively on EU-hosted native cloud platforms (e.g., Mistral, Alf Alpha), while commercial AI applications operate on legally compliant versions of US cloud services. This nuanced stance reflects Europe's balancing act between strict sovereignty and pragmatic openness.

The European Commission’s Strategic Roadmap for Digitalisation and AI in the Energy Sector underscores the urgent need to integrate digital and energy infrastructures to address rising electricity demands driven by digital growth, while enhancing energy efficiency and sustainability. This initiative, part of the broader 2026 Tech Sovereignty Package, aims to balance Europe’s ambitions for technological sovereignty with its climate commitments by promoting AI-enabled energy optimization and sustainable digital infrastructure development.

Central to this sustainability drive, the Cloud and AI Development Act (CADA) seeks to triple Europe’s data center capacity within five to seven years, emphasizing the deployment of highly energy- and resource-efficient facilities that support the green and digital twin transition. The Act introduces a pioneering rating scheme for data centers—covering energy and water efficiency, clean energy use, and waste heat reuse—with the first labels expected in 2027, setting minimum EU energy performance standards to ensure that scaling digital infrastructure does not come at the expense of environmental goals.

However, the ambitious regulatory framework faces pushback from industry groups like the European Data Centre Association (EUDCA), whose secretary general Michael Winterson warns that overly prescriptive energy efficiency directives risk inflating operational costs and deterring investment, potentially undermining the EU’s AI and tech sovereignty ambitions. While directives such as the European Energy Directive have improved transparency through mandatory efficiency reporting, critics argue they have yet to deliver significant energy savings, highlighting the delicate trade-offs between fostering innovation and enforcing sustainability.

The EU Cloud Sovereignty Framework within CADA further exemplifies the complex balancing act by incorporating criteria such as infrastructure location, cybersecurity, and energy use into its four-tier assurance levels, reflecting the intertwined challenges of maintaining technological autonomy while managing energy consumption. As Commission President Ursula von der Leyen emphasized, ensuring control over critical technologies is inseparable from securing stable and sustainable energy grids, underscoring that Europe’s digital sovereignty ambitions must be pursued hand-in-hand with robust energy efficiency and sustainability standards.

Europe’s ambition to develop sovereign AI chips confronts the reality that owning the entire AI stack is unattainable in the near term; instead, the strategic focus is on controlling critical hardware dependencies and governing AI behavior within European society. This 'third way' approach emphasizes embedding compliance, audit trails, and risk thresholds into procurement and market access, leveraging the EU single market to enforce data residency and algorithmic auditability, thereby balancing sovereignty with openness. As noted, “The right reframing for Europe is therefore not ‘can we own the stack?’ but ‘can we control how it behaves within European society and close our most critical hardware dependencies?’”

Despite holding unique upstream semiconductor assets—such as ASML’s global monopoly on extreme ultraviolet lithography and Belgium’s IMEC research—Europe currently exports these advantages rather than harnessing them to build domestic AI chip capacity. This dynamic risks subsidizing foreign AI chip production, particularly in the US, as Christophe Fouquet warns that wafers produced in a European fab would likely be exported to American customers. Consequently, Europe is urged to cultivate a credible domestic customer base through public procurement and investment, learning from China’s state-directed model that created captive demand for Huawei, to avoid becoming a mere supplier rather than a sovereign AI chip ecosystem.

Europe’s AI chip development strategy must distinguish clearly between owning and controlling the AI stack to avoid remaining a declarative ambition rather than an architectural reality. The pragmatic path involves leveraging Europe’s semiconductor design strengths and niche IP—exemplified by companies like Arm—while fostering a fabless ecosystem supported by tax breaks, R&D incentives, and co-investment in design firms. This approach acknowledges the oligopolistic and capital-intensive nature of advanced semiconductor manufacturing dominated by US players, advocating for boosting domestic AI demand and nurturing innovative companies rather than attempting to build politically sensitive mega-fabs.

Balancing innovation with strategic control, Europe is adopting a two-tier AI infrastructure model that separates sovereign government-critical infrastructure from commercial AI workloads, which may run on legally compliant external cloud providers. This nuanced strategy reflects skepticism about regulatory acceptance of certain open-source models in sensitive sectors and recognizes the need to support the 'builders layer'—the innovative companies developing AI models and applications—to drive demand for AI infrastructure. By focusing on engineering excellence and fostering respected products rather than direct confrontation with dominant players like Nvidia, Europe aims to build a sustainable AI ecosystem that harmonizes sovereignty with competitiveness.