Five countries (Cameroon, Kenya, Senegal, South Africa, Tunisia) have fragmented and poorly coordinated digital infrastructures. African universities produce research with tools that are incompatible with one another. Project participants requested local capacities, sovereign data governance, and open platforms to reduce reliance on external tools. DIGITAfrica is developing a distributed and modular architecture based on open and interoperable technical standards, connecting existing resources rather than building a centralized mega-center.

The essentials

  • DIGITAfrica connects five African countries through common standards and free tools, without seeking to centralize resources in a single hub.
  • Its first training sessions in 2026 brought together a significant number of participants, also documenting the fragmentation of equipment, computing, and connectivity across institutions (EPJ Research Infrastructures, 2026).
  • The wager is architectural: scientific sovereignty is built by making existing nodes interoperable, not by replacing them.
  • The project identifies several interdependent conditions, including connectivity, governance, local operational capacity, and recurring funding.

Scientific dependence has a precise geography

The first training sessions in 2026 attracted a significant number of participants and show that gradual scaling is possible if governance is defined from the outset.

A genomics researcher in Nairobi and another in Dakar often work on the same problems. But their data don’t communicate, their servers don’t share resources, and their software doesn’t follow the same standards. When one of them needs computing power for intensive processing, they rent time on an American or European platform, under pricing and contractual terms set elsewhere.

The study published in 2026 in EPJ Research Infrastructures analyzes capacities and fragmentation across five countries, with a detailed institutional survey conducted in Tunisia and documentary research and interviews for the other contexts. The picture shows local gaps in computing and storage as well as reliance on external collaborations to access high-performance computing. The configuration is examined across the five countries studied, but common obstacles are documented only in several countries and are not necessarily present in each one.

South Africa has relatively developed digital infrastructures. Kenya has built a real technology scene around Nairobi. But even in these contexts, research universities operate in silos. Between Tunisia and Senegal, between Cameroon and Kenya, research data circulates poorly and standards diverge. Fragmentation is internal as much as international.

This diagnosis has been known for several years. DIGITAfrica provides a precise architectural response where grand declarations of intent have failed.

Building in network form rather than in pyramid form

DIGITAfrica’s central choice is not to reproduce the mega-center model. Many international scientific cooperation initiatives have followed this logic: a central, powerful facility to which partners access remotely. The result is often displaced dependence rather than built capacity.

DIGITAfrica opts for a distributed-node architecture. Each partner country maintains and develops its own infrastructures, but these are designed from the start to work together. Standards are common, tools are free, and Blueprints define reference architectures, then are tested and refined through pilot deployments. Sovereignty is embedded in technical choices.

This approach has a precedent in other fields. On mobile payments, for example, interoperability between systems has redistributed the balance of power between existing players rather than eliminating them. The principle is the same here: link what exists by imposing common rules, rather than rebuild from scratch.

Concretely, the DIGITAfrica architecture is organized into five deployment levels, supplemented by functional layers. Data management relies notably on FAIR principles, storage services, and interoperability mechanisms. The blueprint provides for a pool of physical and virtual resources orchestrated, to be deployed gradually according to infrastructure tiers. Training and capacity building are integrated into the approach, with activities adapted to local contexts.

77 participants and a method to validate

The 2026 training sessions extend training and demonstration activities already conducted by DIGITAfrica in April 2025. A significant number of participants, from the five partner countries, took part in the first demonstrations and training sessions. The figure is modest relative to continental scale, but it says something important: researchers showed up, institutions played along, and exchanges between countries worked technically.

These first sessions also produced feedback on real obstacles. Connectivity is one of the main bottlenecks, along with computing, skills, governance, and funding. Insufficient bandwidth and network instability hinder access to and sharing of large datasets. The project identifies this point as a critical condition: without robust connectivity, cross-site usage and federated coordination are limited, but architectural interoperability is not merely theoretical either.

The second obstacle is data and computing capacity for artificial intelligence. African universities operate with very different administrative cultures and governance structures. Harmonizing data-sharing policies, intellectual property agreements, and common-access rules between different systems requires negotiations that technical architecture alone cannot resolve. DIGITAfrica has integrated this reality into its roadmap, examining governance models and organizing dialogue with institutional actors and funders.

Infrastructure changes the working conditions of researchers

Access to the DIGITAfrica infrastructure concretely changes the situation of a researcher at a Cameroonian or Senegalese university who currently has no high-performance computing.

The most direct answer concerns data. A public health researcher in Cameroon working on regional epidemiological data currently finds it difficult to cross-reference them with similar data collected in Kenya, due to lack of common formats and sharing protocols. With an interoperable infrastructure, this cross-referencing becomes a standard operation rather than a project in itself.

Training is the second lever. AI is entering African universities before institutions have the resources to train their students to use it rigorously. DIGITAfrica is building an offering in data science and scientific computing education that draws on skills already present in the network, rather than importing programs designed elsewhere for other contexts.

The third lever is publication. African researchers publish in peer-reviewed international journals, but the cost of data processing, rented externally due to lack of local resources, weighs directly on research budgets. Shared and accessible computing capacity reduces this cost, and with it some of the asymmetry that weighs on team competitiveness.

This connects with broader thinking: automated science produces value only if it remains shareable. DIGITAfrica doesn’t automate research, but it creates conditions for results to circulate, accumulate, and be reusable beyond the project that produced them.

Conditions for real scaling up

DIGITAfrica has published a roadmap. It provides for progression through pilots, cross-site experimentation, and regional integration, conditioned by several prerequisites.

Connectivity is treated as an important constraint to be evaluated in cross-site experiments. This depends partly on investments outside the project’s direct control—undersea cables, national broadband policies, peering agreements between operators. The project can design its architecture to take advantage of each improvement; it cannot alone create the network on which it rests.

The project plans dialogue with governments and funders to build stable commitments. A shared research infrastructure cannot operate on limited-duration project funding. Infrastructure sustainability requires viable funding and an economic model beyond the initial funding.

The participation of researchers, institutions, public agencies, and funders is presented as a pillar of sustainability. A network whose users are merely consumers of resources remains fragile. A network whose users contribute data, training modules, and shared protocols becomes an institution. DIGITAfrica designed its first training sessions with this in mind, training trainers rather than delivering lecture-style teaching.

The realization of these conditions would depend on policy choices and investments, and could allow less-equipped universities in the network to access computing resources, data, and advanced training that limited institutional budgets would not otherwise allow them to finance. This scenario remains conditional; it depends on policy choices and investments whose materialization remains to be confirmed. But the architecture that would make it possible is under construction.

The limits of DIGITAfrica

The tension between standardization and diversity also operates at a finer level than that of institutions. Within the same university, research teams don’t have the same needs or equipment levels. A standard retained to be accessible to the least-equipped node mechanically constrains the practices of more advanced teams, who may then seek solutions outside the network. This circumvention weakens overall coherence without the project having direct tools to prevent it, since network membership remains voluntary and shared governance requires a balance between standardization and local autonomy.

The project makes no claim to solve scientific dependence by itself. Three blind spots deserve mention.

The first is scale. The number of partner countries and participants in 2026 activities remains modest relative to ambition; the gap between ambition and current resources is significant. The African Union has 55 member states. Entire regions—Central Africa outside Cameroon, West Africa outside Senegal, all of the Horn of Africa—are absent from the initial network. Future network expansion is mentioned in the roadmap, but its financial and political conditions remain to be specified.

The second blind spot is maintenance. Digital infrastructures age quickly. A network that deploys equipment in 2026 will need to renew or adapt it by 2031. The sustainability of technical resources depends on investment policies that the project can recommend but cannot guarantee.

The third is the tension between standardization and diversity. Adoption of common standards by several different university systems raises compatibility challenges. The most advanced institutions, in South Africa and Tunisia, might find the retained standards too conservative. The least equipped might find them too demanding. Managing this tension without creating hierarchy between nodes remains an important governance challenge.

These limits define the terrain on which the project’s results will be evaluated: the network must produce new and sustainable collaborations, bring benefit to researchers at less-equipped institutions, and enable national institutions to secure sustainable funding after the launch phase.

Sources

  1. Bruno et al., Designing a Pan-African Research Infrastructure for Digital Sciences: DIGITAfrica Vision, Blueprint Methodology, and Roadmap, EPJ Research Infrastructures, 2026, https://link.springer.com/article/10.1007/s41781-026-00184-5