Data sovereignty: Why East Africa must build its own AI future

When a Kenyan bank uses AI for credit scoring or a Rwandan health-technology firm processes patient diagnostics, they are handling what the law deems sensitive protected data.

Sovereign AI, the ability to build AI using a nation’s own infrastructure and governed data is becoming a necessity. It ensures legal compliance with the Office of the Data Protection Commissioner while reducing latency.

Processing data at the edge in local facilities across Nairobi’s financial district or Mombasa’s landing stations eliminates the millisecond delays that can break real-time AI applications like mobile money fraud detection.

However, a common pitfall I see in many local enterprises is the Graphics Processing Unit (GPU) Trap, which is the belief that AI success is simply about buying powerful processors.

In reality, these expensive chips often sit idle because the surrounding infrastructure is insufficient. To move from ambition to execution, East African leaders must address the foundational infrastructure of the digital economy.

AI requires massive amounts of data. Running modern Large Language Models (LLMs) on old storage systems is like installing a race car engine in a vehicle with bicycle tires: the power exists but cannot be used.

There is also a growing cooling challenge. High-density AI workloads generate immense heat that must be managed. As we push toward Green Data Centres in line with regional sustainability agendas and the Bottom-up Economic Transformation Agenda, local firms must adopt energy-efficient solutions to keep their private AI clouds operational.

Security also requires a shift in mindset. Keeping AI in-house doesn't automatically make it safe. Cyber-attacks in the East African Community (EAC) are becoming more sophisticated, often targeting the data pipelines AI relies on.

Forward-thinking firms are moving toward resilience engineering, verifying every user and system every time they access the AI environment.

The effectiveness of any AI system depends heavily on the quality of the data and workflows it operates on. Manual processes, fragmented systems and limited integrations are hurdles. Without clean, organised and accessible data, even the most sophisticated AI systems can produce unreliable results. Firms should therefore digitise and automate core processes before implementing AI.

Building Sovereign AI requires partnership between governments and enterprises. As the EAC moves toward harmonised digital and cybersecurity laws, the winners will be those who invest in strong local foundations. For East African organisations, the opportunity is significant.

The writer is Managing Director, NTT DATA in East Africa