Data infrastructure in the AI era: Power, protection and data justice

The global race for artificial intelligence (AI) dominance is reshaping geopolitics, economics and digital governance. At the heart of this transformation lies data, the raw material that trains and sustains AI systems. The coming years will be defined by the availability of data and the ability to move and process it across borders to enable the development of the best-performing AI systems and models. These dynamics will create new asymmetries of power, influence policy realignments and challenge existing governance models. For children and young people, this evolution carries profound implications for privacy, their rights and inclusion in an increasingly data-driven world. Uncertainty defines the current data landscape. As governments and corporations compete to secure computational and informational advantages, the question of who benefits from AI-driven data extraction and who is left behind remains unsettled. The next few years will test how societies balance innovation with accountability, openness with security, and economic opportunity with ethical responsibility.

Computational capacity and AI as geopolitical assets: AI competition today is not merely  about who has the best AI models, designs and outputs. It is increasingly defined by computational power and capacity: the infrastructure, high-density data centres and land, energy and water required to build, train and run large-scale AI systems. Major economies are scaling up their data centres, cloud platforms and microchip production, turning computational power into a strategic asset and an instrument of geopolitical influence. The three dominant blocks – the United States of America, China and the European Union (EU) – are pursuing different and often divergent models of digital sovereignty. US companies dominate the hyperscale cloud and chip ecosystems (Nvidia, Google, Microsoft, Amazon), while China invests heavily in state-supported AI clusters and supercomputing zones. The EU’s Digital Decade policy programme aims to increase regional computing hubs to reduce dependency on foreign infrastructure. As control over computing power becomes a form of geopolitical leverage, it also shapes who can participate in the digital economy and who remains on its margins.

The rise of mega data centres is a physical manifestation of this AI growth and race for AI dominance. Data centres store and process the vast quantities of information needed to train and deploy AI systems which underpin innovation across many fields that serve humanity, such as health care, education and financial services. They can also create high‐tech jobs and attract investments in complementary industries, such as cloud services, fintech and e-commerce. Yet these benefits are unevenly distributed, and promises of large-scale employment opportunities at a local level may have been (over-)inflated.

Most of the world’s data power still lies in the Global North, and that is unlikely to change soon. In 2024, over 80 per cent of all data centres were located in developed countries and China, with the highest number by far being housed in the United States. Africa, by contrast, hosts less than 1 per cent, while the Middle East will only reach 7.4 per cent of the global capacity in 2029. North America and Western Europe will continue to dominate both the physical data-centre landscape and ownership of global data centres to 2030. While Asia–Pacific records the fastest growth in data centre construction, many of these, with the exception of the data centres located in China, are planned to be funded by North American companies.

The global distribution of computing power will widen existing digital divides. Low- and middle-income countries lack the resources to build advanced AI ecosystems or the infrastructure needed to compete at scale. This computational power imbalance is fast becoming a new measure of inequality and a source of economic and political leverage. In addition, job-creation potential in developing markets is more modest than expected, as many large data-centre projects are highly automated and controlled by foreign operators, offering limited long-term employment for local populations. Without strategic investment in regional cloud and data-processing capabilities, many developing countries risk remaining on the margins of the digital economy, consuming rather than creating the technologies that define their future. For children and families, the potential benefits – such as improved digital access and local services – depend on whether these investments translate into inclusive employment and education opportunities rather than extractive growth.

Building and operating data centres come at a high environmental cost. They require massive amounts of electricity, water and critical minerals such as copper, cobalt and rare earth metals. The International Energy Agency (IEA) projects that global data-centre electricity use could more than double by 2030, reaching about 945 TWh – roughly Japan and Germany’s combined demand. In emerging economies, this growth adds pressure on grids and ecosystems. An Ember Energy study warns that Malaysia’s expansion, for example, could consume 30 per cent of national power demand and drive up carbon emissions. Much of the current data-centre expansion is driven by speculation rather than actual demand, experts claim. At the same time, renewable-energy production is not expanding quickly enough to meet the rising power needs, increasing the risk that countries will rely more heavily on fossil fuels to keep pace. As companies seek cheaper land and power, they will likely expand their operations in lower-income countries where environmental standards may be weaker. This creates a new form of digital extractivism, with the Global South hosting resource-intensive infrastructure but capturing few of its benefits. For children and families, rising energy and water demands can drive up local utility costs, strain essential services and intensify environmental stress in communities already vulnerable to poverty, displacement and climate shocks.

The rise of AI has intensified the demand for global data mobility, yet the rules governing cross-border flows are increasingly fragmented. Since 2017, the number of data-localization laws has more than doubled, as governments seek to protect national sovereignty and citizen privacy. The Organisation for Economic Co-operation and Development (OECD) notes that nearly 100 localization measures were in place across 40 countries by 2023 – and the number is still rising. This fragmentation reflects competing regulatory models. The EU’s General Data Protection Regulation (GDPR) underscores privacy and data‑protection principles; by contrast, regulatory approaches in the United States and some Asian jurisdictions place greater emphasis on enabling data-driven innovation, sometimes at the expense of stricter privacy regimes. The result is a patchwork of incompatible systems that risk creating parallel and fragmented digital economies. This fragmentation may cause regulatory compliance challenges, data breach risks (including breaches of children’s sensitive data), and jurisdictional issues as organizations and companies navigate conflicting privacy rules, age thresholds, profiling restrictions and breach-notification requirements. Furthermore, such fragmentation raises costs, limits innovation and restricts access to diverse datasets that underpin robust AI models.

Implications for children and young people

For children, these dynamics carry far-reaching consequences. As AI systems and education technology platforms increasingly depend on vast and interconnected datasets, children’s data (often collected through schools, learning apps, social networks, and health and social service systems) moves across borders without consistent safeguards. A child’s information may be strongly protected under one country’s data laws and entirely exposed under another. At the same time, strict data localization can block access to global research and technologies that support education, health and protection outcomes. Cross-border data use combined with inconsistent regulations heighten risks of privacy violations, surveillance and algorithmic bias, and may disproportionately harm children. Meanwhile, unequal access to AI and data infrastructure could deepen existing digital divides, leaving children in poorer countries excluded from high-quality learning, participation and future work. In areas hosting data centres where families face rising energy costs and environmental pressures, the need for equitable and sustainable data governance is even more pressing.

The way forward: Data justice for children

Bridging these tensions demands a new paradigm, ‘data justice for children’: a vision of fairness that ensures that children share in the benefits of the data economy while being protected from its harms. Data justice for children means rebalancing who controls and who benefits from data – moving from extraction to fairness, from invisibility to agency. It goes beyond protecting children’s data to ensuring they share the benefits of the digital transformation and have a voice in shaping it. It requires the following:

1. Strengthened international data governance coherence and support for initiatives that harmonize privacy, AI and cross-border data standards to reduce regulatory fragmentation. Children’s rights should be explicitly integrated into multilateral AI agreements, national strategies and data-transfer rules.
2. Innovative approaches to data sharing, such as through federated data systems. Federated data approaches can help balance sovereignty with collaboration. By keeping data within national or institutional boundaries – in schools, hospitals or research centres – these models enable secure cross-border analysis while protecting privacy and building trust.
3. Promoting sustainable digital infrastructure. Governments should adopt ‘green data centre’ policies that tie energy permits and tax incentives to renewable power, water recycling and local employment. Governments in high-income countries should ensure that companies headquartered in their jurisdictions do not exploit weaker environmental protection laws in low-income countries. Developers must engage with host communities, including children, and implement benefit-sharing programmes.
4. Building data literacy and foresight capacity. Equip policymakers, educators and young people with the skills to understand how data systems are evolving and shaping daily life. This can empower the next generation to navigate and influence an AI-driven world responsibly.

Acknowledgements: Thanks to Emma Day, Paula Bruening and Melanie Penagos for input and comments and Yue Wang for data analysis.