The next test for artificial intelligence (AI) is not simply whether it can make economies more productive. It is whether AI can help them produce more value while consuming fewer physical resources: a challenge that will determine whether the digital revolution strengthens or complicates the global sustainability agenda.
Europe is now offering an early answer. A study published in Sustainability, “Adoption of Digital Technologies and Artificial Intelligence: A Panel Data Analysis of Sustainable Economic Performance in European Union Countries,”by Gheorghe Hurduzeu, Ramona Vasilas Pirvu, Laura Nicola-Gavrilă, Cerasela Adriana Luciana Pirvu, Roxana Maria Bădîrcea and Riana Maria Ciobanu, finds that digitalization and AI are linked to higher resource productivity across the EU, although technology delivers its strongest gains when backed by skills and research capacity
Digitalization Helps, but Europe’s Skills Base Does More of the Heavy Lifting
The researchers construct a panel covering the EU-27 from 2015 to 2024, drawing exclusively on Eurostat and World Bank data. Digital adoption is proxied by the share of manufacturing enterprises making e-commerce sales, while the model also includes R&D intensity, tertiary education, government effectiveness and gross fixed capital formation. No missing values were artificially interpolated, and the preferred fixed-effects model uses 256 observations.
Digital adoption shows a positive and statistically significant relationship with resource productivity. The coefficient is modest, which the authors interpret as consistent with gradual efficiency improvements rather than dramatic technological leaps. Digital systems can reduce waste, improve inventory control, coordinate supply chains and make production processes more responsive, but these gains accumulate through countless operational adjustments rather than appearing overnight.
The more striking result concerns education. A one-percentage-point increase in the share of people aged 25–34 with tertiary education is associated with roughly 1.4% higher resource productivity. R&D intensity is also positive and statistically significant, reinforcing the argument that economies need the capacity to absorb, adapt and improve technologies before digitalization can translate into sustainable performance.
Policymakers often frame digital investment, skills policy and green industrial policy as separate portfolios. The study suggests they are deeply interdependent. An economy can buy software, sensors or AI systems relatively quickly; building engineers, researchers, technically capable managers and institutions able to reorganize production around those tools is a much slower process.
AI Looks Promising for Resource Efficiency, But the Evidence Is Still Early
Because harmonized enterprise-level AI adoption data are relatively new, the study treats artificial intelligence separately from its decade-long digital panel. Eurostat provides comparable AI adoption figures only for 2021, 2023 and 2024, leaving the researchers with 79 usable country-year observations. The authors therefore explicitly describe this part of the analysis as exploratory rather than causal.
Even with that limitation, the association is notable. AI adoption has a correlation of 0.51 with resource productivity, and in a random-effects model controlling for human capital, research intensity, investment and governance, the AI coefficient remains positive and significant. A one-percentage-point increase in the share of enterprises using AI is associated with approximately 0.7% higher resource productivity.
The result provides empirical support for the idea that AI can contribute to the green transition through predictive maintenance, production optimization, smarter logistics and better management of energy and materials. But it also cautions against treating AI as a stand-alone sustainability technology. Countries with advanced AI adoption and high resource productivity also tend to possess stronger human-capital stocks, sophisticated research systems and higher-value economic structures.
Efficient, high-income economies may be better positioned to adopt AI in the first place, while technology itself may then reinforce their advantage. The study therefore supports a relationship, not a definitive claim that AI adoption by itself causes greener economic performance.
The Green Transition Has a Material Cost Before It Delivers an Efficiency Dividend
Gross fixed capital formation has a negative short-term association with resource productivity. At first glance, that appears to contradict the idea that investment is essential for modernization. In reality, it highlights a fundamental tension in the transition to greener infrastructure.
Construction, industrial upgrading and infrastructure expansion require large quantities of steel, cement, aggregates and other materials immediately, while their efficiency gains may emerge only years later. A country undergoing an intense investment cycle can therefore temporarily look less resource-efficient even if the new assets ultimately support lower-carbon, higher-productivity growth.
This matters for how governments assess transition performance. Annual resource-productivity indicators may penalize economies undertaking large infrastructure programmes even when those investments are necessary for decarbonization or digital modernization. Policymakers therefore need to distinguish short-term material pressure from longer-term efficiency outcomes rather than assuming that every temporary deterioration signals policy failure.
The study also reveals enormous differences across the Union. Average resource productivity is about €1.90 per kilogram, but values range from roughly €0.30 in Bulgaria to more than €6 in the Netherlands. Such disparities reflect differences in industrial structure, technological maturity, research capacity and the weight of resource-intensive sectors.
For countries already near Europe’s efficiency frontier, the policy challenge is to integrate AI into more sophisticated circular systems, including predictive maintenance, industrial symbiosis and product design that reduces embodied materials. For catching-up economies in Central and Eastern Europe, the more immediate priorities are digital skills, SME support, technology diffusion and stronger links between cohesion funding and resource-efficiency objectives.
Europe’s Twin Transition Will Be Won by Complementarities, Not Technology Alone
The study delivers a broader warning for governments racing to subsidize AI adoption. Digitalization may improve material efficiency, but its returns are conditional. Skills, research systems and economic structure determine whether technology becomes a productivity tool, an environmental tool, or simply another layer of capital expenditure.
Developing countries face an even sharper version of the same challenge because skills shortages, weak research ecosystems and uneven digital infrastructure can limit the productivity return from imported technologies. The implication for the Global South is not to delay AI adoption, but to treat investment in human capabilities, research institutions and digital access as part of the same development strategy.
The research also has important limitations. Its digitalization proxy, manufacturing e-commerce, captures only one dimension of digital transformation, while domestic material consumption excludes resources embodied in imports. Some apparent European “decoupling” may therefore reflect the relocation of resource-intensive production abroad rather than genuine reductions in the material footprint of consumption.
The AI evidence is also constrained by its short time series, and the aggregate country-level analysis cannot reveal which specific applications or sectors deliver the strongest resource savings. Future research will need longer AI datasets, alternative digital indicators such as cloud computing and big-data adoption, sector-level analysis and consumption-based material-footprint measures that account for international supply chains.
Having said that, the strategic message can’t be ignored. The green and digital transitions are not automatically mutually reinforcing simply because policy documents place them side by side. Their intersection depends on whether economies possess the skills, research capacity and institutional depth needed to convert technological adoption into genuine efficiency.
- FIRST PUBLISHED IN:
- Devdiscourse
