Digital transformation in the Shared Socioeconomic Pathways

Digitalisation is not currently represented within the SSP framework. Our quantitative projections of digital transformation levels across countries and time allow the impacts of digitalisation on GHG mitigation and adaptation challenges to be modelled and assessed consistently and comparatively across SSPs. This adds an important new dimension to the characterisation of reference or baseline scenario uncertainty in global development relevant to climate futures. Explicit representation of digital transformation also improves the SSP framework’s wider policy relevance beyond climate, in line with recommendations by O’Neill⁶. Digital transformation is an integral element of Sustainable Development Goals (SDGs)³¹ on innovation, labour markets, and accessible infrastructure (SDG8,9), and a cross-cutting enabler of many other goals, including those on education and cities (SDG4,11).

Here we set out five use cases for how our projections can enrich understanding of GHG mitigation pathways and policies, with an illustrative sixth use case for GHG adaptation (Table 2).

Assessing direct energy and material consumption and resulting GHG emissions from ICT infrastructure

The data centres, networks and devices that comprise ICT infrastructure have a direct energy consumption footprint estimated at 1000 TWh in 2023, equivalent to 4% of global electricity use³². This is projected to increase rapidly in the next 5 years in some locations³³ due to energy-intensive training and inference of generative AI, including large language models. The material needs of ICT infrastructure are small relative to bulk material flows globally, but account for significant shares of certain critical minerals, including rare earths like indium, gallium and germanium³⁴. Our long-term projections of digital transformation levels can be coupled to analyses of the energy and/or material consumption of the continuing expansion of ICT infrastructure across countries and regions under SSP storylines. For example, the marginal effect of digital transformation on energy demand growth observed historically (controlling for other drivers of demand) could be estimated across different time periods and regions. Resulting elasticities of ICT energy demand to deepening digital transformation in turn would enable longer-term projections under different scenario assumptions, including those in the SSPs. Existing SSP variables refine this analysis by contextualising the carbon intensity of electricity and manufacturing activities across different regions (Table 2). Insights on future energy consumption and GHG emissions from ICT infrastructure across countries under SSP storylines, as well as the extent of dependence on rare earth materials, water or recoverable electronic waste, inform policymakers concerned with resource uncertainties and resilience linked with digital trends.

Assessing the indirect impact of specific digital applications (e.g. teleworking) on energy consumption and resulting GHG emissions through changes in behaviour, economic activity or societal functions

Digitalisation impacts energy consumption and GHGs indirectly by influencing or changing household and firm behaviour and so the structure of social and economic activity. Compared to direct impacts, indirect impacts are larger in magnitude, more uncertain, and harder to model for methodological reasons, including difficulties in clearly defining system boundaries^35,36. Impacts can be net energy-reducing (through substitution, efficiency, optimisation) or net energy-increasing (through rebound, induced demand)³⁷. Our projections of the relative levels of digital transformation across different regions provide a means of scaling and extrapolating estimates of indirect impacts from specific digital applications. For example, Hook et al.³⁸ synthesised data showing that teleworking results in a net reduction in overall energy use in transport and buildings sectors ranging from −15% to −0.01% accounting for variation in study design, context, and geography. These estimates of indirect impacts of energy can be combined with SSP variables capturing heterogeneity in relevant adoption conditions across countries (Table 2) to project future indirect impacts of specific digital applications on energy, consistent with SSP storylines. This helps understand their contribution to mitigation goals or the mitigation challenges they pose.

Assessing the indirect impact (enabling effect) of digitalisation as a general-purpose technology on energy consumption and resulting GHG emissions

Econometric models identify relationships between digital transformation and energy demand across the whole economy or in industrial sectors. For example, Briglauer et al.³⁹ found a small but significant negative elasticity of CO₂ emissions with respect to broadband connections in OECD countries over the period 2002–2019. Kopp et al.⁴⁰ found that a 10% increase in firms’ ICT investments was associated with a 0.29% decrease in emissions, with a stronger effect in higher-income countries. These statistical models of the aggregate energy impacts of digitalisation typically control for variation in country size, development stage, trade relationships, and other factors explaining energy demand or GHG emissions. A general finding is that additional digitalisation reduces energy demand at the margins, with a stronger effect in more developed economies. These elasticities identified historically can be integrated with our projected levels of digital transformation across different regions to understand the future economy-wide impact of digitalisation for mitigation goals within different SSPs (Table 2).

Assessing the interaction between digital transformation and the stringency of climate policy required to reach emission reduction targets

By design, SSP reference scenarios assume no additional or new climate policies, and vary widely in GHG emission trends resulting from SSP drivers and other elements. This allows climate policy assumptions to be layered in to identify the stringency required to achieve net-zero or other defined emission reduction or climate stabilisation targets under different SSP uncertainties²⁵. As digitalisation will have significant direct and indirect impacts on energy demand, varying across sector and region, digital transformation will interact with climate policy assessments as both enabler and potential exacerbator. Our projected digital transformation levels allow this to be explored explicitly and quantitatively, including by specifying the relationship between the extent of digitalisation and the carbon pricing levels (or the de-risking of finance) needed to achieve net-zero targets. SSP variables related to governance conditions and effectiveness¹⁵ provide additional context, helping to tailor policy insights to the specific capacities of different regions⁴¹ (Table 2).

Informing global initiatives to ensure universal and equitable access to digital transformation opportunities as part of the SDG agenda

Digital transformation plays a fundamental but complex role in progressing towards SDGs^31,42. Global modelling analyses quantify the synergies and trade-offs of different strategies across the SDGs⁴³ but have not been able to include the interacting effects of digitalisation. Our projected digital transformation levels allow analysis of relationships between SDG indicators and digitalisation across countries under SSP storyline uncertainties. An important example is the digital divide between countries and over time that is shown clearly in our projections (Fig. 3). Modelling the progression of this divide, and how it can be strategically tackled through digital access and infrastructure investments can improve understanding policy capacity (and coherence) to address specific SDG concerns associated with digitalisation (Table 2).

Assessing digital climate services for adaptation planning

Effective climate change adaptation depends not only on traditional resilience strategies but also on the use of digital technologies to improve responsiveness and decision-making. Our example use cases for SSP-consistent projections of digital transformation levels have emphasised impacts on GHG mitigation challenges and opportunities (via energy demand and GHG emissions). However, there are many examples of digital applications both strengthening adaptation planning to climate impacts⁴⁴ (e.g. early warning systems for extreme weather events) but also adversely affecting resilience (e.g. over-dependence on digital infrastructure, digital divides). Our projections can be used to explore how digitalisation influences climate change, adaptive capacity and adaptation planning. For example, empirical studies that show how farmers’ practices are affected by access to accurate weather prediction models or real-time information on market prices for agricultural commodities can be coupled to our projections of how the underlying digital capacities to access and use such tools vary across countries and time (Table 2). This adds a new dimension to SSP-consistent analysis of climate adaptation challenges while emphasising how marked regional variation in digital transformation interacts with the geography of climate impacts. The digital divide potentially adds an additional vulnerability to impact hot spots already subjected to multiple climate stressors⁴⁵.

The digital transformation level projects change in future digital infrastructure, activities, and services, as well as the capacity, including human capital, to support increasing digitalisation. While high digital transformation has the potential to enable sustainable development and net-zero transition pathways, it does not guarantee such outcomes. Digitalisation has its own energy and resource consumption footprint, exacerbated by high carbon intensities of electricity in some futures (e.g. SSP5). Digitalisation also has powerful adverse as well as beneficial impacts across application domains and economic sectors, from governance and societal interactions to industrial processes, jobs and livelihoods.

There are two important limitations and considerations for our SSP-consistent projections of digital transformation. The first concerns uncertainty; the second concerns endogeneity.

Projections based on econometric models using panel data assume that historical relationships observed between variables will hold in the future. While these models can often account for dynamic changes in the short term, uncertainties compound and amplify over the long term. This is particularly the case for digital transformation given its fast-moving innovation cycles, speed of deployment, and potential for surprises²⁷. Our main projections only run to 2050 and emphasise uncertainty around our central estimates.

Some of these uncertainties relevant to digitalisation are represented in the SSP framework and the quantitative projections of drivers, elements and extension variables in the different SSPs. These include rates of technological change and economic growth, and between-country inequalities (Table 1) that indirectly flow through into our projections via their impact on the independent variables in our model.

However, additional uncertainties are specific to digitalisation, particularly those related to breakthroughs in AI⁴⁶ that may result in discontinuous change or even systemic disruption³⁰, rather than the smooth path-dependent trajectories projected by our historically calibrated panel models using future GDP, population, and R&D intensity trends.

Against these smooth trajectories, the effects of AI, generative AI, or other digital advances not captured in historical relationships can be explored using sensitivity analysis, sector-specific projections, or further what-if scenario analysis—all of which can be nested within the SSP framework’s long-term variation of macro uncertainties relating to the economy, demography, urbanisation, and technological change.

These issues of uncertainty are arguably less fundamental than issues of endogeneity. Our approach treats digital transformation as an internally consistent outcome of socioeconomic development already characterised by different SSPs. However, it is not only an outcome. Digital transformation is also a driver or amplifier of socioeconomic uncertainty, as our analysis of the digital divide compounding income inequality shows. At the same time, digitalisation can contribute to productivity gains and accelerate economic growth.

This two-way endogenous relationship—digitalisation as both driver and outcome of change—was also emphasised in our expert workshop mapping of linkages between digital transformation and SSP elements (Supplementary Table 3).

As our approach takes the SSP framework as given—both as a set of future narratives and as derived demographic and economic variables used in our modelling—we do not capture the feedback effects of digital transformation on social and economic development. The same limitation applies (by design) to climate impacts on the socioeconomic development pathways, which determine GHG emissions but do not co-evolve with the resulting changes in climate.

Other SSP extensions, such as gender inequality and the rule of law, also face this endogeneity problem as they both impact and are impacted by socioeconomic development. As with our projections of digital transformation, the value of these SSP extensions is in making important uncertainties explicit in order to open up further avenues for policy-relevant analysis (see our use cases). What is distinctive about digital transformation, however, is the speed and magnitude of its potential disruptiveness.

As an example, digital transformation as a driver of change may undermine governance institutions and political agency through misinformation, unchecked market power of tech companies, and social polarisation under assumptions of weak global oversight³. Generative AI could turbocharge this challenge to the governance landscape necessary for concerted action on global commons problems like climate change, highlighting risks similar to the tragedy of the commons⁴⁷. To some extent, this is captured implicitly in the SSP3 storyline of fragmentation, divergence, and weakened global institutions. But could AI accelerate and amplify this effect to the point at which it destabilises the fundamental socioeconomic assumptions and relationships on which the SSP framework is built? This eventuality is explored by Carlsen et al.³⁰ in the case of a breakthrough on AGI²⁸. Resulting systemic disruption could require a reimagining of a wholly new scenario architecture for understanding future uncertainties relevant to climate goals.