Navigating Digital Sovereignty in the Era of Great Power Rivalry

Envisioning AI Infrastructure as Digital Public Infrastructure

In recent years, Artificial Intelligence and “Digital Public Infrastructure” (DPI), the two technological developments that have emerged as key policy imperatives and generated significant interest in global discussions on technology sovereignty and economic development. Current digital economy is characterised by BigTech capture and technological dependency on a handful of mostly US-based companies. As the global race to dominate the artificial intelligence landscape intensifies, the governments and corporations have geared to drive substantial investments in underlying digital infrastructure, that is central to gaining AI supremacy. In the mid of february 2025, a new report EuroStack was launched to secure European Union's (EU) digital sovereignty to ensure an independent, open, sustainable digital future of the EU. The report is touted to be the most comprehensive and ambitious contribution so far to the policy discourse on European digital sovereignty and the strategic role of digital public infrastructures. This is the latest and one of the crucial policy directions among many digital tech initiatives around the world that has broadened the DPI scope. IndiaStack is considered as one of the major sources of inspiration for other DPI initiatives around the world. The other notable DPI initiatives being launched are from Brazil, Ethiopia, Estonia etc. The DPI initiatives are largely from the global south and confined to provide scalable and interoperable technologies and systems that provide citizen/public services via digital technology platforms. Though there are various examples of Digital Public Infrastructures (DPIs) yet there is no common definition to understand what it really entails. DPI was defined at the G20 Summit of August 2023 as “a set of shared digital systems that should be secure and interoperable, and can be built on open standards and specifications to deliver and provide equitable access to public and/or private services at societal scale". Open Future Foundation has proposed an expanded conception of digital public infrastructure, moving beyond software and governance to encompass both the hardware and social frameworks that form the critical foundations for effective technological rollout. It also calls for open, democratic discourse to address the diverse goals these infrastructures can serve. As digital public infrastructure spreads globally, national and region-specific adaptations and innovative conceptualisations are emerging. These span a spectrum from digital commons to its integration into industrial policy and its interplay with artificial intelligence.

BigTech Control Over Compute and Digital Infrastructure

The perceived potential of new advancements in generative AI, from anticipating productivity gains to accelerating scientific breakthroughs, has sparked widespread enthusiasm. However, this rapid deployment of advanced AI models hinges on two foundational pillars: compute (computational power) and cloud infrastructure. These pillars not only enable AI innovation but are increasingly concentrated under the control of a handful of U.S. tech giants, raising concerns about power imbalances and systemic risks.

Compute, a critical layer in the AI supply chain, comprises a complex stack of interdependent components:

• Hardware: Specialized chips like GPUs (Graphic Processing Units) and CPUs form the backbone of AI systems, enabling parallel processing for training complex models.
• Software: Tools like Nvidia’s CUDA programming framework optimize chip performance, acting as gatekeepers to hardware capabilities.
• Infrastructure: Data centers—with their servers, cooling systems, and cabling—aggregate hardware into scalable computational powerhouses.

This infrastructure is not neutral. Its design and accessibility shape who can participate in AI development, privileging entities with the resources to commandeer vast computational resources.

The digital infrastructure underpinning AI is dominated by a near-oligopoly of the U.S.-based corporations such as Amazon (AWS), Microsoft (Azure), and Google (Cloud), which collectively control 70% of the global cloud market. Their dominance extends to physical infrastructure, including over half of the world’s undersea internet cables (when Meta is included). These firms have leveraged their monopolies over cloud services and compute resources to exert unparalleled influence across economic, social, and political spheres.

The ensuing AI boom has also deepened existing power asymmetries. According to a report from Oxford University, only 32 countries host large-scale AI compute hubs, while the United States and China control more than 90% of global AI data centers. Over 150 countries lack access to advanced computing infrastructure altogether. Remarkably, Harvard’s AI lab alone possesses more compute power than all African-owned facilities combined.

These hubs, vast data centers powered by high-performance GPUs, have become indispensable for AI development, scientific research, and digital services. The divide they create is not only technological but also geopolitical. Training advanced large language models such as GPT-4 requires immense computational and infrastructural resources, accessible almost exclusively through hyperscale cloud providers. Even rising competitors, including major Chinese tech firms, struggle to overcome the barriers to challenging the US BigTech hegemony. Meanwhile, the environmental costs of these facilities, from soaring energy consumption to heavy water usage, remain largely under-regulated, as corporations prioritise rapid scalability over sustainability.

In essence, AI’s promise is inextricably tied to a system where computational power and infrastructure are controlled by a few Big Tech gatekeepers.

In essence, AI’s promise is inextricably tied to a system where computational power and infrastructure are controlled by a few BigTech gatekeepers. This concentration risks stifling real innovation, exacerbating environmental harm, and deepening societal dependence on entities whose priorities are driven by profit, not public interest.

AI Supremacy and Digital Nationalism Eroding Multilateral Governance

In an era defined by geopolitical rivalry and the urgent quest for digital sovereignty, the Global South faces mounting pressure to build resilient DPI amid a fractured global landscape. While initiatives like the ’50 in 5′ campaign, backed by multilateral organizations and philanthropies, aim to accelerate DPI adoption in 50 countries by 2029, the absence of cohesive leadership and major powers like the U.S. and China pursuing national AI goals threatens to undermine these efforts. Both nations prioritise competing technological paradigms: the U.S. champions market-driven, private-sector dominance, while China exports state-aligned infrastructure through initiatives like the Digital Silk Road. This rivalry leaves the Global South navigating a polarised ecosystem, where dependency on foreign technologies risks eroding local control over data and AI governance. Meanwhile, the promise of open, collaborative “digital commons”, exemplified by projects like Wikipedia and Apache, offers an alternative vision of shared infrastructure. Yet, these decentralised models struggle to scale without robust support, as geopolitical tensions and the lack of aligned incentives among powerful states fragment resources and standards. The result is a precarious balancing act for Global South nations, caught between great-power agendas and the urgent need to secure equitable, sovereign digital futures. As countries prioritise technological sovereignty and data security, divergent regulatory frameworks and technical standards are emerging, complicating cross-border interoperability.

Global South nations are increasingly pressured to align with competing tech blocs, fragmenting DPI development. For example, U.S.-led initiatives like the Clean Network campaign discourage partnerships with Chinese firms, limiting affordable options for countries like Indonesia, which faces costly overhauls to comply with Western data norms. This contestation forces Global South governments into a lose-lose scenario: choosing between China’s cost-effective but opaque ecosystems and the West’s pricier, regulation-heavy alternatives. For example, Brazil’s GOV.BR digital platform, designed to integrate AI-driven public services, has faced delays as U.S. cloud providers resist data localization demands, while Chinese alternatives trigger scrutiny over security risks. The absence of neutral, multilateral frameworks leaves these nations navigating a fractured landscape, where their DPI priorities, such as inclusivity and ethical AI, are sidelined by great-power agendas. This dynamic not only stifles innovation but also deepens global inequities, as the Global South remains a battleground for technological dominance rather than a partner in shaping equitable digital futures.

The absence of neutral, multilateral frameworks leaves these nations navigating a fractured landscape, where their DPI priorities—such as inclusivity and ethical AI—are sidelined by great-power agendas.

Digital public infrastructure (DPI) initiatives in the Global South face mounting challenges in achieving AI sovereignty, as they grapple with resource constraints, technological dependency, and inequitable global power dynamics. Many countries lack the funding, skilled workforce, and computational infrastructure to develop homegrown AI systems, forcing reliance on foreign technologies that often come with strings attached. For instance, Ethiopia’s nascent AI strategy struggles with limited access to cloud computing and high-performance hardware, pushing it to depend on partnerships with foreign firms like Huawei or Western tech giants, which retain control over critical data and algorithms. Similarly, Kenya’s digital ID system, Huduma Namba, has faced setbacks due to biases in outsourced biometric technologies and public distrust over opaque data-sharing agreements with foreign vendors. These dependencies risk perpetuating ‘digital colonialism’, where Global South nations cede control over their data ecosystems to external actors, undermining their ability to shape AI governance frameworks that align with local needs. Compounding this, stringent intellectual property regimes and restrictive licensing terms imposed by Global North corporations further marginalise efforts to build sovereign AI capabilities.

Can EuroStack be the Model for Global Digital Public Infrastructure?

The recently launched EuroStack report warns that AI and digital tech’s future is being dictated by the U.S.-China duopoly controlling 90% of advanced semiconductors and 75% of cloud infrastructure. This dependency, EuroStack argues, leaves the EU and the Global South vulnerable to supply chain disruptions and algorithmic imperialism. EuroStack’s most radical proposition is a “fair compute” initiative, redistributing 20% of EU cloud capacity to partner nations in Africa, Asia, and Latin America. This would enable regions lacking data centers to train AI models on local languages and priorities, a direct challenge to the current paradigm where Global South nations lease computational power from foreign hyperscalers, often forfeiting control over data and innovation. However, it remains to be seen how this will be implemented in reality.

EuroStack’s most radical proposition is a “fair compute” initiative, redistributing 20% of EU cloud capacity to partner nations in Africa, Asia, and Latin America.

It proposes binding standards for energy-efficient data centers, mandating renewable energy use, and taxing high-intensity AI training runs. Crucially, it reimagines computational infrastructure itself: incentivising modular, repairable hardware over disposable server farms and prioritising public cloud systems that share resources across institutions. But EuroStack risks replicating past power imbalances. Its ‘partnerships’ with the Global South remain framed through European strategic interests. Yet, by centering sustainability and sovereignty in the same agenda, it could provide a model for the global south countries to come up with a "global south stack" policy vision. This could provide a crucial counterweight to dominant AI narratives, where technological progress isn’t built on ecological sacrifice or digital colonialism.

A Call for A Commons-Led DPI

Building a digital ecosystem that prioritises people and the planet demands collective action: governments and communities must unite to dismantle the unchecked dominance of tech giants, reclaim shared knowledge and data, and redistribute power and profits concentrated in corporate hands. Recently a policy proposal was mooted by a coalition of researchers to offer a roadmap to build a digital stack for people and the planet. Central to this vision is the creation of a ‘public-led digital stack’ grounded in open, ecological infrastructure, universal platforms, and democratically governed services, alongside robust public education to counter Big Tech’s manipulative narratives. The proposal advocates for "a democratic, publicly governed digital stack consisting of: 1) digital infrastructure as a service, covering training, processing, and solution development, managed by non-profit, international democratic consortia; 2) universal platforms, including search engines and foundational AI models, treated as shared commons and overseen by new public institutions that include both state and civil society representation; and 3) a public marketplace where companies can offer computing services free from lock-ins. To ensure sustained demand, governments should procure services from this marketplace and terminate contracts with Big Tech."

DPI governance should be transparent and accountable. Regular audits of DPIs’ social, economic, and environmental impacts should be undertaken and made public. While international cooperation is vital, countries can take immediate steps by enacting regulations to foster sovereign digital foundations and dismantle reliance on extractive corporate systems. Critically, these efforts must be interconnected; standalone digital public infrastructure (DPI) projects risk becoming mere extensions of Big Tech’s control if not anchored in a holistic, publicly owned ecosystem. Only by integrating material infrastructure, ethical governance, and civic empowerment can societies forge a sustainable alternative to today’s exploitative digital paradigm, one that serves collective needs rather than private profit. Such a publicly led digital stack can safeguard the digital sovereignty of states striving for autonomy from those that control critical digital infrastructure. This is especially important at a time when powerful nations are competing for global AI supremacy. It can also offer communities a viable path to reduce structural dependencies.

This article is a revised version of the blog article that I wrote in April 2025.