Romania Is Last in the EU on Company AI Adoption — 5.2%. Employees Are Last Too

5.2%: Romania, Last in the EU on Companies

Eurostat published the results of its latest survey on AI technology use in EU enterprises on December 11, 2025. The definition is broad — text mining, speech recognition, text/image/video generation, image recognition, machine learning, and AI-based process automation — applied to companies with at least 10 employees or self-employed people, across NACE sections C-J, L-N and 95.1 (manufacturing, construction, trade, transport, IT, real estate, professional activities, repair services).

The EU average climbed to 20.0% in 2025, up from 13.5% in 2024 and 8.1% in 2023 — nearly tripling in two years. Denmark (42.0%), Finland (37.8%) and Sweden (35.0%) lead the ranking. At the other end: Romania (5.2%), Poland (8.4%) and Bulgaria (8.5%). Romania sits at barely two-thirds of the next-lowest country and at roughly a quarter of the EU average.

The figure excludes companies under 10 employees, so it says nothing about freelancers or micro-businesses — the segment where AI might actually be more present, informally. What Eurostat measures is AI adoption as part of a company's declared activity with real structure, not one employee's individual experiments.

Employees Are Last Too

The natural assumption is that employees got there first — that they're already using ChatGPT or Claude on their own, even if the company has no official policy. A separate Eurostat survey, published December 16, 2025, shows otherwise: in 2025, only 32.7% of people aged 16-74 in the EU had used any generative AI tool — 25.1% for personal use, 15.1% at work, 9.4% for education. The lowest shares in the entire EU: Romania (17.8%), Italy (19.9%) and Bulgaria (22.5%).

Romania is therefore last in the EU on both measures — companies and individuals alike. That directly contradicts a press release widely cited over the past month, from Bittnet Training and LLPA, claiming that "2 out of 3 urban Romanians" already use AI daily. The gap between 66% and 17.8% is too large to be explained by sampling alone — the two likely measure different things: any interaction ever, versus actual use in the last three months, on a nationally representative sample. It's the kind of figure mismatch that should raise a flag: when a statistic sounds far more optimistic than everything else around it, check the primary source before repeating it.

Romania doesn't fare better on the infrastructure behind AI either: only 24.9% of Romanian companies used paid cloud services in 2025, against a 52.7% EU average (Eurostat, February 3, 2026) — Finland, Italy and Malta all clear 75%. The cloud gap (roughly 2x below average) is real, but proportionally smaller than the AI gap (roughly 4x below average) — so infrastructure explains only part of the total AI gap.

The Capability Exists, in Pockets

If technical talent were the problem, it would be hard to explain why Romania is home to companies like UiPath, Bitdefender and Endava — names with internationally recognized AI and automation expertise. The skills exist in the market; the gap is in implementation, at scale, at the average company.

PwC published the Romanian edition of its CEO Survey on February 25, 2026 (79 Romanian CEOs surveyed, part of a global survey now in its 29th edition). More than half of CEOs report minimal AI use in core company processes, and non-use reaches two-thirds in some areas. But in a few specific functions, Romania beats the Central and Eastern Europe average: product development (20% versus 14% CEE average), demand generation and marketing (19% versus 15%), and strategic direction (19% versus 13%). 23% of CEOs say AI has already cut costs at their company.

The CEE AI Index 2026, published by AI Chamber together with The Recursive Media (June 3, 2026; 11 countries, 33 indicators), ranks Romania 7th out of 11 regionally, with a score of 41.1 points on the "Environment" pillar (ecosystem maturity) — behind Estonia (73), Poland (65) and Lithuania (64), but ahead of Bulgaria (29) and Croatia (27). Romania leads the region on internet speed and has a national AI strategy (2024-2027), backed by more than €64 million through the HRIA program (2025-2029). Ana Marija Kostanic, editor-in-chief of The Recursive Media, sums up the tension: "Romania has invested seriously in the strategic layer — public policy, infrastructure, European partnerships. The adoption gap is real, but there's also meaningful momentum to close it."

The conclusion from these three independent sources: technical and strategic capability exists, at least partially. What's missing is the mechanism that turns that capability into everyday use at the mid-sized company — the one with 20, 50 or 150 employees, not the tech giants where AI is already built into the business model.

What This Means for Your Company

McKinsey estimates AI could unlock €280-700 billion in Central Europe — 6-15% of the region's net turnover — if the adoption gap closed. That's a potential, not a guarantee: nearly 60% of the region's economy sits in sectors where AI is hardest to scale (manufacturing, construction, retail), where only 17-18% of companies have reached the scaling stage, versus 43% in tech.

For an ordinary company, the findings above translate into a concrete sequence, not a large digital-transformation project:

• Take an honest inventory: what your employees are already using individually (ChatGPT, Claude, Copilot), even informally — start from what already exists, not from zero.
• Pick one repetitive, measurable process (invoicing, support triage, reporting, lead qualification) — not "AI across the whole company" in one move.
• Define clearly what data the system can access and who owns the outcome — the point where most deployments get stuck.
• Measure time, cost and resolution rate on the chosen process for a few weeks before extending to the next one.

It's the same logic laid out in a broader guide to operational AI agents: real scaling starts from one well-defined process, not a generic pilot. Romania already has the technology and the talent available in the market; what's missing is the number of companies that have already gone through the sequence above and can show the result.

MassAI editorial policy: our analyses combine primary sources and independent publications. When we cite estimates or commercial reports, we explicitly state the source and context.

Sources: ↗ Eurostat — AI adoption in companies · ↗ Eurostat — generative AI use by individuals · ↗ Eurostat — paid cloud services · ↗ McKinsey — Capturing Central Europe's AI opportunity · ↗ Forbes.ro — CEE AI Index 2026 · ↗ PwC — CEO Survey Romania 2026