Why Digital Transformation Is Failing to Deliver Its Promised Carbon Efficiency Gains

The rapid expansion of digital infrastructure may be delivering an environmental outcome at odds with economic ambitions. New research finds that greater digital connectivity is associated with declining carbon emission efficiency, while more intensive information and communication technology (ICT) activity can deepen the environmental burden.

The study, "Digitalization, Energy Use, and Carbon Emission Efficiency: Testing the Digital Paradox," published in the journal Energies, examines how digital development affects environmental performance across 53 economies. Authored by Fatma Mabrouk, Ahmed A. Mashaal, Muhammad Amjed Iqbal and Muhammad Sohail Amjad Makhdum, the research investigates whether technological expansion delivers efficiency gains sufficient to compensate for its growing energy requirements.

Drawing on economic, technological, institutional and environmental data covering 2002–2022, the researchers identify a statistically significant negative relationship between digital infrastructure and carbon emission efficiency. Their findings challenge the assumption that expanding digital access automatically supports decarbonization, pointing instead to a structural tension between technological progress and the physical resources required to sustain it.

When digital efficiency gains collide with rising energy demand

Digital technologies can improve production planning, optimize energy management, reduce transaction costs and support more efficient resource allocation. However, their operation depends on electricity-intensive infrastructure, including communication networks, servers, digital equipment and data-processing facilities. As digital services expand, additional energy consumption can offset the efficiencies these technologies generate elsewhere in the economy.

The researchers describe this contradiction as the digital paradox, a situation in which technological improvements coexist with increasing environmental pressure. They identify three possible mechanisms: infrastructure expansion increases electricity and material requirements; lower operating costs encourage additional consumption through rebound effects; and digital investment potentially competes with resources intended for environmentally oriented innovation.

To investigate these relationships, the study measures carbon emission efficiency by assessing how effectively economies generate economic output while using labor, capital and energy and accounting for carbon dioxide emissions. Its methodology incorporates emissions as an undesirable outcome, allowing researchers to evaluate environmental performance alongside productive efficiency.

Across the 53 economies, the average carbon emission efficiency score was 0.6946, indicating considerable distance from the estimated efficiency benchmark. More importantly, the baseline analysis identifies a negative digital infrastructure coefficient of −0.1677, statistically significant at the 1% level. The result indicates that economies with greater digital infrastructure development tend to exhibit weaker carbon efficiency after accounting for the factors included in the model.

The negative relationship remains significant under alternative statistical methods, reinforcing the consistency of the finding. Nevertheless, the researchers acknowledge that the analysis cannot fully eliminate reverse causality or omitted-variable influences. The evidence therefore supports a persistent negative association rather than definitive proof that digital expansion independently causes carbon efficiency to deteriorate.

ICT intensity exposes a deeper environmental vulnerability

Digital infrastructure reflects access and connectivity through mobile subscriptions, internet use, fixed broadband and fixed telephone services. ICT intensity instead captures the scale of economic involvement in ICT goods exports, imports and service exports.

The difference reveals that expanding access and expanding ICT-related economic activity do not necessarily produce identical environmental outcomes. An economy can become more digitally connected without developing an equally intensive ICT production or trade structure. On the other hand, greater ICT activity may introduce substantial additional energy and material requirements alongside productivity improvements.

When the researchers examine both dimensions together, digital infrastructure and ICT intensity show significant negative relationships with carbon efficiency. Their interaction produces a coefficient of −0.6864, indicating that the adverse association between connectivity and environmental performance becomes stronger as ICT-related economic activity increases.

The estimated marginal relationship between digital infrastructure and carbon efficiency also deteriorates across ICT intensity levels. At the 25th percentile, the coefficient stands at −0.1570, compared with −0.1743 at the median and −0.2373 at the 75th percentile. These findings suggest that the environmental pressures accompanying digital expansion become more pronounced in economies with greater ICT-related activity.

The results raise questions about development strategies that measure digital progress primarily through connectivity expansion, technological adoption and ICT-sector growth. The researchers argue that environmental performance depends partly on whether expanding infrastructure and ICT activities are supported by efficient equipment, cleaner electricity and sustainable production practices.

Green innovation offers limited relief, while governance emerges as a decisive factor

Innovation is frequently presented as the mechanism through which technological development can generate environmental improvements. Digital systems can facilitate knowledge exchange, research collaboration and technological diffusion, potentially accelerating the adoption of cleaner production methods. However, the study identifies a more complicated relationship between digital infrastructure, innovation and carbon efficiency.

The researchers find a positive association between their innovation indicator and carbon emission efficiency, represented by a coefficient of 0.0089. Yet digital infrastructure is negatively associated with innovation, with a coefficient of −0.0820. One possible explanation advanced by the authors is that expanding digital infrastructure may compete with environmentally oriented research for limited financial, technological and human resources.

Their mediation analysis reveals an especially important limitation. Innovation explains only 0.44% of the estimated total negative association between digital infrastructure and carbon efficiency. Although the indirect relationship is statistically significant, its magnitude is extremely small, leaving the underlying negative relationship largely unchanged.

The innovation finding also requires careful interpretation. The researchers use total resident patent applications as a proxy for green innovation because comparable environmental patent data were unavailable across the full sample and study period. Since general patent applications include technologies unrelated to environmental protection, the results cannot establish how specifically green inventions influence the relationship.

Institutional quality produces a considerably stronger positive association with carbon efficiency, recording a coefficient of 0.4047. The study combines indicators covering government effectiveness, regulatory quality, rule of law, political stability, accountability and corruption control, suggesting that effective governance provides an important foundation for better environmental performance.

Industrial structure moves in the opposite direction, with a negative coefficient of −0.0598. The researchers associate this finding with the environmental demands of resource-intensive industrial activity, reinforcing the importance of cleaner production and industrial upgrading. Their analysis also identifies a U-shaped relationship between economic development and carbon efficiency, suggesting that efficiency may deteriorate during earlier stages of economic expansion before improving as economic structures and technologies advance.

Together, these results position environmental performance as an outcome shaped by technological development, economic structure and institutional conditions. Digitalization alone cannot substitute for effective regulation, cleaner industrial systems or sustained environmental innovation.

The policy challenge: Building digital economies without multiplying their carbon burden

The findings introduce an important consideration for governments pursuing simultaneous digital transformation and environmental sustainability. Expanding broadband networks and ICT-related industries may support economic modernization, but their environmental performance cannot be assumed from technological advancement alone. Digital investment policies consequently require closer integration with energy planning, industrial development and environmental governance.

The researchers recommend prioritizing energy-efficient data centers, communication networks and digital equipment while expanding renewable energy use. They also advocate cleaner electricity, improved ICT-sector production practices and environmental objectives embedded directly within digital development strategies. Such measures would address the physical energy requirements accompanying wider digital connectivity.

For developing economies, the findings highlight the importance of coordinating digital infrastructure investment with institutional capacity and cleaner industrial development. The study includes both developed and developing economies, although it does not provide a separate causal assessment of how individual development groups experience digital expansion. Its conclusions therefore support policy caution rather than a uniform prescription across countries.

Stronger environmental regulation, transparent reporting and regular monitoring also feature prominently in the researchers' recommendations. Governments are encouraged to assess digital policies through energy consumption, emissions and environmental innovation, rather than relying exclusively on connectivity, infrastructure availability and investment indicators.

The evidence nevertheless has boundaries. National-level data can conceal important differences between firms and industries, while the digital indicators do not comprehensively measure data-center activity, hardware production or digital technology use across sectors. The study also ends in 2022, leaving subsequent technological developments outside its empirical coverage.

Future investigations could address these gaps through industry-specific analysis, environmental patent indicators, more detailed digital infrastructure measures and stronger causal identification methods. The authors also recommend examining whether countries with different energy intensities experience distinct relationships between technological expansion and carbon efficiency.