Why Europe’s hidden AI advantage lies in the application layer

AI adoption is the defining business and geopolitical challenge of the decade. Early movers are gaining a compounding advantage, widening the gap before others can respond.

But is there an AI bubble? The answer depends on one critical question: how quickly AI can deliver real value through applications. To understand that, let’s consider Nvidia’s strategy.

Jensen Huang uses the metaphor of a five-layer cake to describe today’s AI landscape: energy, chips, data centres, models and applications. He argues that the application layer is both the most important one and the one that is still largely missing. Nvidia is now investing massively to support companies building this layer, because how fast it delivers value will determine whether the bubble bursts or fuels one of the most significant economic transformations ever seen.

The real question is whether the application layer can mature fast enough to create lasting economic value.

Energy is energy. AI chips remain largely application-agnostic. Specialized inference chips are emerging, but GPUs are still the de facto general-purpose solution. Data centres can support a wide variety of workloads. Foundation models can be specialized, but they are remarkably, and sometimes disappointingly, good at performing general-purpose tasks. Training hyper-specialized foundation models with much less data rarely pays off.

AI applications are different. They are incredibly diverse, because the experience expected by humans varies widely depending on the task at hand. Building the right tools at this layer requires an intimate understanding of the actual users. Building AI for engineering teams, for instance, means understanding not just how engineers work today, but who they are and what they could become with better tools.

That is why the application layer is much more fragmented than the other layers. It is also much harder to brute-force success, even with deep capital and world-class researchers. The application layer of the AI cake will not be uniform. It will be a patchwork of flavours.

Understanding users this deeply leads to a crucial insight: the goal of AI in the workplace should not simply be to maximize what machines can do. It should be to elevate what humans can achieve. In complex engineering environments, AI should help experts explore more options, understand complex trade-offs, and make better-informed decisions, rather than simply accelerating execution.

Productivity alone is an incomplete metric. Real value emerges when AI strengthens the human dimensions of work: purpose, mastery and accomplishment.

Yet most organizations struggle with this shift — adoption is widespread, but meaningful impact remains limited, largely due to poor integration with human workflows.

Take product engineering and design. AI systems will soon be able to handle an increasingly large share of the design process for complex systems, whether a car, a spacecraft, a mobile phone or a nuclear power plant. This means that the throughput of engineering teams could increase dramatically, augmenting their ability to explore new ideas and concepts, driving an unprecedented acceleration of progress.

What’s missing today is the right interfaces between humans and AI. We need to redefine the role of human engineers and develop the right interaction patterns for engineers to work effectively with AI systems. For engineers, these experiences are highly context-dependent.

This challenge is more complex than many observers assume. Anthropomorphic paradigms like “virtual employees” can be seductive, but AI systems fail differently from humans and react at different speeds. Genuine productivity emerges from human-AI collaboration, where both contribute their strengths.

Debate often focuses on model competitiveness and large-scale infrastructure, where Europe faces a well-documented gap. But a more decisive factor is now emerging: the quality of the human and AI interface in the application layer. This remains a widely unsolved problem, and solving it will require nuance, creativity and deep domain understanding.

Europe holds underestimated expertise in precisely this area. Its long track record in developing sophisticated software for complex technical domains has produced deep capabilities in human-centred interfaces for high-stakes environments, where AI must support expert judgement rather than override it. In product engineering and computer-assisted design, two European companies, Dassault Systèmes and Siemens, were among the world leaders of the last major digitalization wave. Today, in several areas of AI for engineering, European companies are again building strong positions.

The numbers support this optimism: Europe’s software-engineering talent pool includes a per-capita concentration of AI experts 30% higher than in the US and almost three times as high as in China. Europe also ranks second globally in AI research publications, and saw AI funding hit a record $21.8 billion in 2025, up 58% in a single year.

In the application layer, the usual handicap of European companies, which are often less funded at inception, may become less critical. Because this layer is more fragmented, less capital-intensive and less amenable to brute-force investment, it may offer exactly the kind of opportunity in which European companies can compete and win.

But this opportunity will not materialize automatically. Europe must bring AI-powered specialized applications to market fast, learn from real adoption patterns, and iterate relentlessly. The winners in the application layer will not simply be those with the biggest models or the deepest infrastructure. They will be those who understand the user best, integrate most effectively into real workflows, and design systems that genuinely expand human capability.

This demands a clear commitment: investing in human-AI collaboration, maintaining human accountability, and rewarding better decision-making — not just faster execution.

So, is there an AI bubble? Perhaps: after all, there is speculation, excess capital and inflated expectations. But the more important question is whether the application layer can translate this moment into durable value. If it can, then what looks like a bubble today may instead be remembered as the investment surge that enabled a profound transformation of the economy.

And if that is where value will ultimately be created, then Europe has a meaningful hand to play.