Uganda's Aeonian Project: Africa's First Sovereign AI Factory Is Being Built on the Nile
June 2026 | Infrastructure & AI
A $1.2 billion AI facility is currently taking shape inside the Karuma Hydropower Plant on the River Nile. For tech professionals tracking Africa's infrastructure trajectory, the Aeonian Project is worth paying close attention to - not just as a headline, but as a meaningful architectural shift in how AI compute is provisioned on the continent.
The Problem It's Solving
The statistic that frames this project is stark: approximately 98% of African data is currently processed outside the continent. That means African researchers, startups, and enterprises building AI systems are almost entirely dependent on foreign cloud providers - AWS, Azure, GCP - for the compute they need. The latency, cost, currency risk, and data sovereignty implications of that dependency are significant, and they've been largely accepted as the cost of doing business in Africa's tech ecosystem.
The Aeonian Project is a direct attempt to change that calculus.
What's Actually Being Built
The facility is structured as a 100MW hyperscale Tier-4 Plus hybrid off-grid green energy Data & High-Performance Computing Centre (DHPC), divided into six 15MW AI modules (KRM1-KRM6) plus 10MW dedicated to supercomputing, totalling 100MW at full build-out.
At the core is USIO, a sovereign supercomputer built in partnership with NVIDIA, AI infrastructure firm MDCS.AI, and Belgian automation company Automation NV. USIO runs on NVIDIA's Blackwell GPU platform - the same architecture powering frontier AI workloads globally - making it genuinely competitive hardware rather than a mid-tier compromise.
Rollout timeline:
- H2 2026 - Phase 1: 15MW AI module + USIO supercomputer goes live
- 2027 - Full 100MW capacity reached across all six modules
- 2028 - Sequential completion of remaining modules ensuring modular autonomy
The Engineering Choices Are Deliberate
What makes the Aeonian Project technically interesting is how tightly the infrastructure design is tied to its physical location.
Power: The facility draws on surplus pre-transmission electricity from Karuma's 600MW output - up to 100MW of renewable hydropower that would otherwise be underutilised. This solves one of AI infrastructure's most pressing problems (energy cost and availability) using existing capacity.
Cooling: Rather than energy-intensive mechanical cooling, the facility uses natural Nile river water. Combined with modular heat-reuse technologies, this makes Aeonian one of the few AI data centres globally with a genuinely low environmental footprint by design, not just by offset.
Connectivity: Two leased dark fibre optic cables link the facility to Kampala, and onward via a 2,500 km fibre network to submarine cables in Kenya and Tanzania - plugging Uganda directly into the global internet backbone.
Why Sovereignty Matters for Developers
For engineers and researchers building on this infrastructure, the sovereignty angle isn't just political - it has practical consequences:
Local language model training. One of the persistent gaps in African AI is the lack of models trained on local languages and African contextual data. With sovereign compute on the continent, institutions can train models on Luganda, Kiswahili, and other regional languages without routing sensitive datasets through foreign jurisdictions.
Data residency compliance. As African nations develop their own data protection frameworks, the ability to guarantee that data never leaves the continent becomes a compliance requirement, not just a preference.
Lower latency for regional applications. Applications in healthcare, agriculture, and fintech that depend on real-time inference benefit directly from compute that's geographically close to their users.
The Broader Context
The Aeonian Project isn't happening in isolation. It's part of a coordinated regional push:
- The African Development Bank and UNDP launched an AI 10 Billion Initiative at the 2026 Nairobi AI Forum, targeting a $1 trillion GDP impact by 2035.
- East Africa is actively pursuing an AI sovereignty agenda, prioritising systems built on local data and hosted on regional infrastructure.
- Uganda's Ministry of ICT is finalising a National AI and Emerging Technologies Strategy this month, providing the policy framework that will govern how facilities like Aeonian are used.
International backing includes Germany's GIZ, Finland's HAUS, the EU Development Fund, and other European development agencies - signalling that this is not a speculative venture but an infrastructure bet with multilateral support.
What This Means in Practice
For developers in Uganda and East Africa, the operational launch of Phase 1 in H2 2026 represents something concrete: access to NVIDIA Blackwell-class compute without routing workloads through US or European cloud regions. For startups, that's a meaningful cost and latency improvement. For researchers, it's the possibility of building models that actually reflect African data distributions.
The comparison MDCS.AI co-founder Niels Van Rees draws is pointed: "Data and AI will define economic opportunities in the coming decades. This facility positions Africa to lead in innovation rather than follow." Whether that ambition materialises depends on execution - but the technical and financial foundations being laid at Karuma are more serious than anything the continent has attempted in AI infrastructure before.
The Nile has powered Uganda for decades. In 2026, it's starting to power something else entirely.
