EU businesses have pushed artificial intelligence adoption to nearly 20 percent. The numbers look promising on paper. But scratch the surface and the picture fractures.
According to Eurostat data released in late 2025, 19.95 percent of enterprises with at least 10 employees used at least one AI technology that year. The figure climbed 6.47 percentage points from 2024. Large companies led at 55.03 percent. Medium-sized ones reached 30.36 percent. Small firms lagged at just 17 percent. The gap between large and small widened to 38 points.
And. Regional differences stand out sharply. Denmark hit 42 percent. Nordic neighbors followed close behind. Romania managed only 5.2 percent. These spreads reveal a continent still operating in silos rather than a unified market.
Generative AI use tells a similar story. The European Investment Bank Investment Survey found 37 percent of EU firms deployed it. That matched the US rate. Yet American companies applied the technology across more business functions. Eighty-one percent of US adopters integrated AI into at least two processes. The EU share stood at 55 percent. Europeans focused on internal operations and marketing. Americans pushed further into customer service, human resources and beyond.
Manufacturing offered one bright spot. Forty-eight percent of EU manufacturers used big data and AI. The US figure sat at 28 percent. Robotics adoption followed suit with EU manufacturers at 55 percent versus 36 percent stateside. But services and infrastructure told the opposite tale. US firms led in Internet of Things deployment.
The The Next Web analysis published today cuts to the heart of the matter. Headline adoption gains hide a stubborn truth. Capital does not flow easily. Skills remain scarce. And the single market exists more in theory than practice. European firms still buy most of their AI capabilities from American cloud providers. Those three dominant US clouds control about 70 percent of the EU market. European providers claim just 15 percent.
Mario Draghi laid out the diagnosis last year in his landmark report. Successive European Councils answered with statements and pilot programs. Little changed on the ground. The productivity gap with the US, estimated at 70 percent per capita in some analyses, traces back two decades. Technology shortfalls explain roughly two-thirds of it.
Skills shortages top the list of barriers. Half of small and medium-sized enterprises cite them as a primary obstacle, per OECD findings referenced in recent analyses. Maintenance costs worry 40 percent. Hardware availability and regulatory questions follow. An Ipsos report from March 2026 titled Making AI work for Europe outlines six questions that block decisions. Is the technology relevant to my work? Does it deliver clear value? What about risks around cybersecurity, privacy and accuracy? Do the rules make sense? Can I trust the outputs? And do I have the capability to make it work?
Workers express particular concern. Sixty-six percent in one Eurobarometer poll believed AI would destroy more jobs than it created. Trust in AI outputs hovered around 46 percent in related surveys. These human factors matter as much as technical ones.
Yet regulation receives more attention than it perhaps deserves. The EU AI Act enters full force in phases through 2026 and 2027. Compliance burdens exist. They hit smaller firms hardest. But the Act does not represent the primary obstacle. Implementation delays announced recently, pushing some high-risk system deadlines to 2027 or 2028, reflect the complexity. Still, the deeper issues predate the legislation.
Venture capital tells its own story. US AI startups drew $194 billion in 2025. European counterparts secured $15.8 billion. That disparity limits company formation and scaling. It leaves promising research in labs rather than in products. Arthur Mensch, chief executive of Mistral, has called for Europe to own and operate its own AI infrastructure. His company recently took on significant debt to build a major data center in Paris.
Data readiness compounds the challenge. Only 22 percent of organizations possess the foundations needed to scale advanced AI use cases, according to a recent Strategy& report. Without clean, accessible, well-governed data, models deliver limited impact. Many pilots stall. Return on investment stays elusive. McKinsey has noted that while global AI usage reaches high levels, only a small fraction of organizations report material earnings before interest and taxes gains.
The European Commission responded with its AI Continent Action Plan in April 2025. The plan targets trustworthy AI, large-scale computing infrastructure, better data access and skills development. The Apply AI Strategy launched later that year focuses on small and medium-sized enterprises. It promotes an AI-first mindset across sectors while weighing benefits against risks. Regulatory sandboxes under the AI Act aim to test applications in controlled settings.
But. These initiatives face the same structural headwinds. Fragmentation across 27 member states complicates scaling. Different regulatory interpretations add friction. Talent mobility suffers from incomplete labor market integration. Capital markets remain national in character despite years of promises.
Recent coverage echoes these tensions. A Brookings Institution piece from May 2026 highlights how AI productivity gains concentrate among firms with strong digital skills, quality data, capable management and reliable cloud access. That uneven diffusion risks widening existing divides. An Amazon Web Services report cited in March coverage noted a 26 percent rise in European AI investment. Confidence grows. Yet only 31 percent of businesses maintain a formal AI strategy. Four and a half million firms tried the technology for the first time last year. Scaling remains the test.
Fortune magazine in March examined Europe’s position in the next wave of AI. The piece pointed to industrial strengths in factories and talent pools but noted persistent gaps in commercialization and venture funding. US startups captured 74 percent of global AI venture capital in 2024 while Europe took 12 percent. The funding shortfall in information and communications technology and cloud infrastructure runs into the trillions over time.
So what now? Policymakers talk of completing the single market for capital and services. They discuss tax incentives for AI investment and pan-European venture funds. Education systems need reform to produce more AI-literate graduates and mid-career professionals. Public procurement could drive demand for homegrown solutions. Cross-border data spaces might reduce reliance on foreign clouds.
Business leaders must act too. They cannot wait for perfect conditions. Successful adopters treat AI as a business transformation exercise, not a technology project. They invest in change management. They redesign processes around human-AI collaboration. They measure outcomes rigorously. Schneider Electric offers one model. The French industrial firm runs about 100 AI applications. Savings projections reach €400 million annually. Leadership aims for every product and function to feel the impact eventually.
The stakes run high. Productivity growth has slowed across advanced economies. AI represents one of the few plausible levers for acceleration. Europe possesses scientific excellence, regulatory credibility and a large internal market. It lacks execution at scale. The gap between Nordic leaders and southern or eastern laggards cannot persist if the bloc wants to remain competitive.
Recent PwC predictions for 2026 warn that many agentic AI projects may be canceled due to costs, unclear value or weak controls. More than half of global CEOs report no significant financial return from AI spending so far. The technology has outpaced mobile phone adoption in speed of uptake. Translation into sustained business value proves far harder.
Europe stands at a crossroads. Adoption rates have caught up in some headline measures. The ability to extract real economic benefit has not. Fixing capital allocation, skills pipelines and market integration would do more than any regulation. Without those changes, the 20 percent figure may represent a ceiling rather than a foundation. The next few years will show whether European firms can move beyond experimentation to genuine transformation. Or whether they will continue buying American tools while hoping the structural problems solve themselves.