Africa must invest in digital infrastructure, computing capacity and regional connectivity if it is to exercise greater control over its data and capture more of the economic value created by artificial intelligence (AI), the Chair of the Governing Board of the National Information Technology Agency (NITA), Estelle Akofio-Sowah, has said.
Speaking at the 11th MOBEX Africa Innovation Conference, Ms Akofio-Sowah said Africa’s digital economy was expanding across payments, public services, healthcare, education and business, but the continent needed to look beyond adopting technology to who owned and controlled the systems powering it.
“Digital sovereignty isn’t just about data location. It is about having the local capacity to manage, secure and monetise our digital assets effectively,” she said.
She said digital sovereignty required investment in fibre networks, resilient data centres, reliable power and local computing capacity, supported by clear governance arrangements.
However, she cautioned against interpreting digital sovereignty as a call for countries to isolate themselves from global technology markets. Instead, African countries should build regional capacity, share infrastructure and strengthen connections across borders to avoid unnecessary duplication and reduce costs.
Ms Akofio-Sowah cited open-access fibre networks as one way to expand connectivity by allowing multiple service providers to use shared infrastructure rather than build separate networks.
She said the same principle could support government digitalisation, with critical public infrastructure connected through shared networks that improve access and service delivery.
But infrastructure investment alone would not guarantee success. She said commercial viability depended on policy, public procurement and private-sector demand working together.
“Policy, public sector procurement and private demand must align to drive active utilisation,” she said.
She also drew attention to recurring fibre cuts, which force operators to spend repeatedly on repairs and can disrupt services. Data centres, she said, could not deliver their full value without reliable network connections.
Investors, therefore, needed to view fibre, data centres, power and computing capacity as parts of one connected value chain rather than separate investments.
For Africa, she said, the priority was to build the infrastructure and skills needed to participate more fully in the digital economy, attract investment and retain more value locally.
Africa must move from AI adoption to control
Nehemiah Yelu Attigah, Founder and Chief Executive Officer of Hatua Tech Ltd, said Africa’s growing use of artificial intelligence did not necessarily mean the continent controlled the technology or the economic value it generated.
“Adoption is not the same as control,” he said in his keynote address.
He said Africa increasingly used advanced AI tools but remained dependent on external providers for much of the infrastructure, models, rules and financing behind them.
Mr Attigah identified five areas that would determine Africa’s influence over AI: data, computing capacity, models, capital and rules.
These, he said, raised fundamental questions about who generated and controlled data, who owned AI models and intellectual property, who financed innovation and who set the standards governing the technology.
He pointed to African-developed applications as examples of how AI could serve local needs and reflect the continent’s languages, institutions and history.
Kaya AI, for instance, supports learning in African languages, including Ewe. Odigro uses AI to search Ghanaian parliamentary records dating back to the pre-independence period, while Veridara applies AI to legal intelligence grounded in Ghanaian laws and court cases. Akofina uses AI to support a game featuring African historical figures.
Mr Attigah said such initiatives demonstrated how AI could help preserve democratic records, improve access to professional knowledge and make African history and culture more accessible.
However, he stressed the need for safeguards, particularly where AI is used in legal and public-interest settings. Legal tools should provide citations and should not replace professional judgment, while systems that search parliamentary records must distinguish analysis from official documents and identify the sources supporting their responses.
He also cautioned against assuming that every African country needed to build its own large language model. Countries could use open-source technology, global cloud services and cross-border partnerships while developing the expertise to audit systems, negotiate fair terms and change providers when necessary.
The objective, he said, should be to ensure that Africa could make informed decisions about the technologies it uses rather than remain a passive consumer of systems designed and controlled elsewhere.
