Faster Internet, No Cleaner Air: Satellite Evidence Exposes the Limits of Digital Growth

Rapid digital transformation is reshaping economic activity, but its environmental benefits may be considerably less automatic than policymakers assume. Evidence from Uzbekistan suggests that even dramatic improvements in internet infrastructure do not necessarily translate into cleaner air, challenging a widely promoted connection between digital connectivity and environmental progress.

A study titled "Is Faster Internet Associated with Cleaner Air? Satellite Evidence on Digital Infrastructure and Nitrogen Dioxide in Uzbekistan and Central Asia," published in the journal Sustainability, examines this relationship through satellite pollution observations and internet performance measurements. Authored by Sardorbek Kodirov, Elnorakhon A. Muminova, Go'zal Xankeldiyeva, Sherzod Dehkanov and Dilnoza Khudoyberganova, the research investigates whether improved network quality corresponds with reductions in nitrogen dioxide (NO₂), a pollutant associated with combustion.

Examining 161 Uzbek districts between 2019 and 2025, alongside 42 regions across Central Asia, the researchers found no measurable reduction in NO₂ associated with faster mobile internet. Despite a 7.4-fold increase in mobile download speeds across comparable districts, the estimated pollution response remained close to zero. The findings raise questions about whether digital infrastructure alone can deliver the environmental gains frequently associated with technological modernisation.

The green digital dividend faces a reality check

Digitalisation is frequently linked to cleaner economic growth through several potential mechanisms. Remote working can reduce commuting, digital coordination can improve logistics, and connected production systems can help businesses manage energy and resources more efficiently. Online public services may also replace physical journeys, potentially reducing transport-related emissions.

However, these benefits compete with opposing pressures. Expanding telecommunications infrastructure consumes energy, while improved connectivity can stimulate businesses, employment, trade and additional economic activity. Greater efficiency does not necessarily mean lower overall consumption when technological improvements encourage further production and demand.

Previous research has often reported favourable environmental relationships, particularly in studies examining China's provinces and cities. However, many investigations measure digital development through composite indicators incorporating telecommunications, employment, financial services and broader economic activity. Such measures can make it difficult to separate the influence of connectivity from other development processes.

Environmental measurements introduce another complication. Local carbon emissions are frequently estimated by distributing national totals across smaller geographic areas using population, road networks or industrial locations. These methods support emissions modelling but may provide limited independent evidence of how pollution changes within individual districts.

Uzbekistan offered an opportunity to investigate the relationship using more direct measurements. Its Digital Uzbekistan-2030 strategy accompanied rapid telecommunications improvements, while road transport, industry, power generation and heating remained relevant sources of combustion-related pollution.

The researchers thus focused on network quality rather than an aggregated measure of digital development. Their approach examined whether actual improvements in internet performance corresponded with changes in independently observed atmospheric NO₂, allowing a narrower assessment of the proposed environmental benefit.

Sevenfold connectivity growth delivers no detectable pollution reduction

The scale of Uzbekistan's connectivity expansion was substantial. Median district mobile download speed increased from 7.8 megabits per second in 2019 to 54.4 in 2025. Median fixed-broadband speed rose from 6.8 to 113.0 megabits per second, while mobile latency declined from 102.4 to 32.6 milliseconds.

Across 158 districts with mobile measurements in both endpoint years, the average increase in download speed was approximately 7.4-fold. Despite this transformation, the research identified virtually no relationship between faster mobile networks and reduced atmospheric NO₂.

The principal statistical estimate produced an elasticity of 0.001, meaning the estimated pollution response to changing network speeds was extremely small. Its 95% confidence interval extended from −0.0129 to 0.0150, encompassing both modest pollution decreases and increases.

Translated into practical terms, doubling mobile download speed was associated with a possible NO₂ change between approximately −0.89% and +1.04%. Across the entire observed network expansion, the estimated association was bounded between roughly −2.6% and +3.0%.

Crucially, the researchers went beyond reporting a statistically insignificant result. An equivalence test supported the conclusion that the relationship was small, providing stronger evidence against substantial changes in NO₂ associated with mobile-speed improvements.

The finding remained broadly consistent after accounting for population, weather conditions, regional trends and differences in satellite processing. Alternative analyses using quarterly observations and network measurements from consistently observed locations also failed to identify a significant relationship.

Researchers additionally examined whether digital improvements produced stronger benefits in urban districts, the Tashkent metropolitan area or particular seasons. None of the three hypotheses received statistical support after adjustment for multiple comparisons.

The absence of a detectable seasonal difference is particularly relevant because heating contributes heavily to cold-weather NO₂ pollution. Connectivity improvements alone would not necessarily alter this major combustion source.

The wider Central Asian analysis similarly found no significant association between mobile download speeds and NO₂. However, a secondary finding linked lower mobile latency with higher pollution at the regional level when consistently reprocessed satellite observations were used.

Such an association could reflect economic expansion accompanying improved connectivity, although the authors caution against causal interpretation. The latency result was not a primary prespecified hypothesis and requires independent investigation.

Why pollution measurements can change the digital economy narrative

The research exposes weaknesses in environmental datasets commonly used to evaluate local economic development. The researchers compared satellite-observed NO₂ with geographically allocated carbon emissions from the Emissions Database for Global Atmospheric Research. For several combustion-related sectors, estimated district emissions largely reproduced national growth patterns rather than revealing distinct local changes.

Between 2019 and 2024, 41.6% of districts recorded estimated emissions growth from residential and commercial buildings within one percentage point of national growth. For road transport, the corresponding proportion was 20.6%.

Satellite NO₂ measurements displayed considerably greater geographic variation. Only 0.6% of districts showed pollution changes within one percentage point of the national pattern.

This contrast suggests that some geographically allocated emissions estimates contain limited information about district-specific developments. Statistical relationships involving such data may partly reflect allocation methods rather than independently observed environmental changes.

Satellite observations offer an alternative, but they introduce technical challenges of their own. A processing change implemented in 2021 increased reported NO₂ values, with larger adjustments in more polluted locations. Without correction, this discontinuity could distort estimated relationships between connectivity and pollution.

Comparisons with consistently reprocessed satellite observations demonstrated the importance of accounting for such changes. The study consequently offers a methodological lesson: environmental conclusions depend not only on statistical techniques but also on whether the underlying measurements capture the local changes being investigated.

Importantly, NO₂ and carbon dioxide measure different environmental outcomes. The absence of a measurable relationship between network quality and NO₂ cannot establish that digitalisation has no effect on greenhouse-gas emissions.

Digital infrastructure needs environmental action beyond connectivity

The findings challenge the assumption that telecommunications investment should automatically count as an environmental intervention. Faster networks can support economic development, but their presence alone does not demonstrate reductions in combustion-related pollution.

For governments pursuing digital transformation, the distinction between infrastructure provision and technological use is essential. Broadband speed indicates the quality of available networks, not whether businesses are optimising energy consumption, households are replacing journeys with online services, or industries are adopting cleaner production processes.

The researchers identify several explanations consistent with the results. Digital efficiency gains may be too limited to influence total pollution, economic expansion may offset potential savings, or structural changes may take longer than the seven-year observation period to emerge.

These possibilities remain unresolved because the analysis does not directly measure district-level digital adoption, application use or the economic mechanisms connecting connectivity with emissions. Its observational design also cannot establish causation.

Nevertheless, the findings suggest a clearer division of policy responsibilities. Telecommunications programmes can pursue connectivity and economic objectives, while air-quality improvements require interventions addressing combustion in transport, heating, electricity generation and industry.

For development agencies, satellite observations offer opportunities to monitor environmental outcomes geographically rather than relying exclusively on modelled emissions. Such monitoring requires careful attention to atmospheric conditions, satellite processing and local data quality.

The research also points toward a more targeted agenda for future investigation. Better information on internet adoption, business activity, transport behaviour and energy use could establish whether particular digital applications generate environmental benefits that network-speed measurements cannot reveal.

Longer observation periods and independently documented telecommunications upgrades could further clarify whether effects emerge gradually or differ across economic sectors.

The key finding does not invalidate digital infrastructure investment or dismiss its potential environmental value. Instead, it establishes an important boundary around what connectivity improvements alone can demonstrate.

Uzbekistan's experience shows that technological progress and cleaner air cannot be treated as interchangeable outcomes. Digital networks may enable changes that reduce pollution, but achieving those reductions depends on how technology is adopted, which economic activities it transforms and whether direct action addresses the combustion sources responsible for deteriorating air quality.