Critical Technologies Race: How China, US and EU Compare in Radical Innovation (Bruegel 2025)

Why Radical Novelties Matter for Critical Technologies

In the escalating global competition for technological supremacy, not all innovation is created equal. While incremental improvements drive steady progress, it is radical novelties — patents representing genuinely new inventions never before documented — that reshape industries, shift geopolitical power, and determine which nations lead the next technological era. A landmark 2025 Bruegel working paper by García-Herrero, Krystyanczuk, and Schindowski provides the most comprehensive analysis yet of how China, the United States, and the European Union perform in creating and absorbing these breakthrough innovations across three critical technology domains.

The study examines artificial intelligence, semiconductors, and quantum computing — the foundational technologies that will underpin everything from national defence and economic productivity to healthcare and energy systems over the coming decades. By deploying large language models to analyse millions of patent filings, the researchers identify which patents represent true radical novelty and measure how quickly those breakthroughs spread across borders. The findings reveal a troubling picture for Europe and important strategic insights for policymakers worldwide. For organizations navigating the shifting technology landscape, understanding these dynamics is essential to making informed decisions about technology strategy and competitive positioning.

How Bruegel Identifies Breakthrough Critical Technology Patents

What distinguishes this study from conventional patent analysis is its innovative methodology. Traditional patent counting treats all filings equally, whether they describe a minor improvement to an existing process or a fundamentally new approach. The Bruegel team deploys large language models (LLMs) to evaluate individual patents and classify them as either incremental improvements or radical novelties — inventions describing something that has never been patented before.

The methodology draws on patent data from major international offices including the European Patent Office (EPO), the USPTO, and China’s CNIPA. For each patent, the LLM assesses whether the claimed invention represents a genuinely novel concept or merely an iteration on prior art. This AI-driven classification allows the researchers to identify the truly frontier innovations within the vast ocean of annual patent filings.

Once radical novelties are identified, the study traces their replication patterns — measuring how quickly a breakthrough patent from one jurisdiction gets replicated or built upon in another. This spillover analysis reveals not just who creates innovation, but who absorbs it fastest, and which technology corridors carry the most cross-border knowledge transfer. The dual-lens approach of novelty identification plus spillover mapping produces a far richer picture of technology competition than traditional metrics like R&D spending or total patent counts.

US Dominance in Quantum Computing Innovation

Among the three critical technologies examined, the United States demonstrates its most decisive advantage in quantum computing. American institutions — spanning government laboratories, major technology companies, and research universities — produce a significantly higher volume of radical novelty patents in quantum than either China or the EU. This leadership extends across quantum hardware (qubit technologies, error correction, and cryogenic systems), quantum algorithms, and quantum communication protocols.

The study finds that China and the EU show broadly comparable performance in quantum radical novelties, both trailing substantially behind the US. While China has made headline-grabbing advances in quantum communication and specific hardware configurations, these have not yet translated into a systematic lead in radical quantum patenting. The EU, despite notable research programmes at institutions such as Quantum Flagship and leading universities, has failed to convert its academic expertise into the kind of frontier patent output that defines technology leadership.

For quantum computing, the implications are stark. The United States has effectively established a structural advantage in the most foundational computing paradigm of the coming decades. Nations and blocs that fall behind in quantum radical novelties today risk permanent strategic dependency as quantum technologies mature and find commercial application in cryptography, materials science, drug discovery, and financial modelling.

Critical Technologies in AI: US Leadership and China’s Strategic Subfields

In artificial intelligence, the picture is more nuanced than in quantum computing. The United States slightly outperforms China in overall AI radical novelty production, but the margin is narrow and varies significantly by subfield. Most notably, the US shows clear dominance in generative AI — the technology behind large language models, image generation, and multimodal systems that have captured global attention since 2023.

China, however, demonstrates considerable strength in specific AI subfields. The study highlights Chinese leadership in areas such as aerial vehicle technology (autonomous drones and unmanned systems), where Chinese researchers produce more radical novelties than their American and European counterparts. This pattern of specialised strength mirrors China’s broader industrial strategy: rather than competing across every frontier simultaneously, China concentrates radical innovation effort in domains aligned with its military and economic priorities.

The EU’s position in AI is the most concerning of the three regions. European institutions produce fewer radical AI patents than either the US or China, and critically, the EU shows the slowest replication speed for AI innovations originating in other regions. While the EU has world-class AI research talent at universities and public labs, the ecosystem for translating that research into patented frontier inventions — and then into commercial products — remains weaker than in the US and China. This aligns with a well-documented European challenge: strength in basic research but weakness in commercialisation and scaling.

Semiconductor Innovation: China’s Breadth vs High-Value Design

The semiconductor findings present perhaps the most strategically significant and counterintuitive results in the study. China dominates a larger number of semiconductor subfields than either the US or the EU when measured by radical novelty patent counts. This breadth of semiconductor innovation reflects China’s massive industrial push to build self-sufficient chip capabilities, driven by the Made in China 2025 initiative and subsequent government investment programmes totalling hundreds of billions of dollars.

However, the study reveals a critical caveat: China does not dominate in semiconductor design, which represents the highest value-added segment of the chip industry. Semiconductor design — the creation of novel chip architectures, instruction sets, and IP cores — is where the largest share of industry profit and strategic leverage resides. Companies that control chip design (like ARM, Qualcomm, NVIDIA, and AMD) capture far more value than those focused on fabrication or testing alone. In this highest-stakes arena, the US and EU maintain meaningful advantages.

This bifurcation carries profound implications. China’s breadth in semiconductor radical novelties means it is building genuine capability across the full chip stack — from materials and lithography to packaging and testing. Yet its relative weakness in design suggests that pure volume of patenting does not automatically translate into dominance over the industry’s most strategic chokepoints. For EU policymakers, this is a reminder that targeted investment in high-value semiconductor design capabilities may offer better returns than attempting to match China’s sheer scale of patenting across all subfields.

Cross-Border Technology Spillovers and Replication Patterns

The Bruegel study’s analysis of technology spillovers — how quickly radical novelties from one region are replicated in another — reveals patterns that should alarm European policymakers and inform global strategy. The central finding is that AI radical novelties travel fastest across borders, followed by semiconductors and then quantum computing. This speed differential likely reflects the relatively open nature of AI research (with extensive preprint sharing and open-source models) compared to the more closely guarded semiconductor manufacturing processes and quantum hardware specifications.

Within this cross-border flow, the most striking pattern involves the EU: Europe is the slowest of the three regions at replicating radical novelties originating in the US and China. When American or Chinese researchers patent a breakthrough in any of the three critical technologies, European institutions take longer to produce comparable or derivative innovations. Conversely, when European researchers create a radical novelty, the US and China replicate it relatively quickly. This asymmetry — slow to absorb, quickly absorbed — represents a structural disadvantage that compounds over time. Those looking to understand how knowledge flows across complex organisations may find value in exploring interactive analyses of major industry reports.

Between the US and China, the replication of radical novelties moves surprisingly fast in both directions. Despite the geopolitical tensions, decoupling rhetoric, and active export control regimes, breakthrough innovations continue to flow between the world’s two largest technology powers with remarkable speed. This finding has major implications for the effectiveness of technology containment strategies.

Why Export Controls Fail to Stop Critical Technology Diffusion

One of the study’s most policy-relevant findings concerns the limited effectiveness of export controls in preventing technology diffusion. The US has implemented increasingly stringent restrictions on semiconductor technology exports to China, including advanced chipmaking equipment, design software, and high-end chips themselves. These controls are intended to slow China’s technological advancement in strategically sensitive areas.

Yet the Bruegel data shows that radical novelties continue to replicate quickly between the US and China across all three technology domains. This persistence of spillovers despite export controls suggests several mechanisms at work. First, the fundamental knowledge embedded in radical patents is often published and accessible through patent databases, academic papers, and conference proceedings — information flows that cannot be easily controlled. Second, the global mobility of researchers and engineers creates human channels for knowledge transfer that transcend trade barriers.

Third, and perhaps most importantly, the nature of radical innovation means that even when specific technologies are restricted, the underlying scientific principles enable independent re-creation. If a Chinese researcher understands the principle behind an American quantum error correction technique, they can develop their own implementation without needing to import the original. This suggests that policymakers relying primarily on export controls to maintain technological advantage may need to recalibrate their strategies — combining restrictions with accelerated domestic innovation to stay ahead rather than simply trying to slow competitors down.

Europe’s Critical Technologies Gap: Diagnosis and Solutions

The Bruegel study paints a sobering picture of Europe’s position in the global critical technology race. The EU’s challenges are twofold: it produces fewer radical novelties than the US or China across all three technology domains, and it is slower at absorbing breakthrough innovations from other regions. This combination of lower output and slower absorption creates a widening competitiveness gap that threatens Europe’s economic sovereignty and strategic autonomy.

Several structural factors explain this gap. Europe’s venture capital ecosystem, while growing, remains significantly smaller than those in the US and China, making it harder for frontier technology startups to scale rapidly. The EU’s regulatory environment, while designed to protect citizens and promote fair competition, can create compliance overhead that slows the commercialisation of new technologies. And Europe’s fragmented market — with different national innovation systems, procurement processes, and industrial strategies — makes it difficult to concentrate resources at the scale needed for frontier technology development.

However, the study also reveals areas of opportunity. Europe’s strength in fundamental research provides a foundation for radical innovation. The EU’s deep expertise in specific semiconductor design capabilities, its world-class quantum research programs, and its leadership in responsible AI development represent assets that, if properly leveraged through coordinated industrial policy, could reverse the current trajectory. Gaining insight into how competitive dynamics play out across industries can help organisations refine their strategies — explore more interactive analyses of global industry reports.

Policy Roadmap for Critical Technology Competitiveness

Based on the Bruegel findings, a clear policy roadmap emerges for each major actor. For the European Union, the priorities are urgent and interconnected. First, Europe must dramatically increase frontier R&D funding specifically targeted at radical innovation in AI, semiconductors, and quantum computing. The current levels of investment, while substantial in absolute terms, are insufficient to close the gap with US and Chinese output. Second, the EU must accelerate technology absorption through strengthened public-private partnerships, streamlined procurement processes, and dedicated knowledge transfer mechanisms that speed the journey from international breakthrough to European implementation.

Third, Europe needs a comprehensive talent strategy that both retains European researchers and attracts global experts. The brain drain of top AI and quantum researchers to US companies and institutions remains a critical bottleneck. Fourth, industrial policy must focus on building complete innovation ecosystems — from university research through startup creation to scaleup funding and market access — rather than funding isolated projects. The OECD’s innovation policy recommendations provide a framework for such ecosystem-level interventions.

For the United States, the study reinforces the importance of sustained R&D investment and talent pipelines, while cautioning that export controls alone cannot maintain technological advantage. The US must combine selective restrictions with aggressive domestic innovation acceleration. For China, the findings suggest that its breadth-first strategy in semiconductors is bearing fruit in terms of radical patent output, but capturing the highest-value design capabilities will require a different approach focused on fundamental research and IP creation rather than volume alone.

Strategic Implications for Global Technology Leadership

The Bruegel study on radical novelties in critical technologies arrives at a pivotal moment in global technology competition. As nations increase investment, tighten controls, and compete for talent, understanding who actually creates breakthrough innovations — and how those innovations spread — becomes essential for strategic planning at every level, from government policy to corporate R&D allocation to individual career decisions.

The core message is clear: radical innovation capability is concentrating in the US and China, with the EU falling further behind in both production and absorption. AI spillovers move fast and are difficult to contain. Quantum computing leadership remains firmly American. And in semiconductors, sheer patent volume matters less than controlling the high-value design chokepoints that define industry power.

For European leaders, the window for course correction is narrowing. The technology architectures being built today in AI, quantum, and semiconductors will define strategic dependencies for decades. Acting now with coordinated investment, talent attraction, and industrial policy can still close the gap. Waiting risks permanent subordination in the technologies that will shape the twenty-first century economy. For professionals seeking to stay ahead of these shifts, engaging deeply with the original research — rather than relying on summaries — is the foundation of informed strategy. The Bruegel working paper provides an essential starting point for any serious assessment of where global technology leadership is heading.