Why Nigeria’s AI future depends on breaking government data silos

Eight major government agencies hold some of Nigeria’s most valuable datasets on citizens. But these databases remain siloed, with little interoperability or data sharing. Industry experts say that unless this changes, Nigeria’s push to become a leading artificial intelligence power in Africa could struggle before it truly begins.

AI systems need large amounts of high-quality data to work well. When data is stored in separate databases using different formats and standards, it becomes harder to build AI tools that can improve services such as healthcare, education, tax collection, and identity verification.

The fragmentation spans some of the country’s most important databases. The National Identity Management Commission (NIMC) manages the National Identification Number (NIN) database, while the Central Bank of Nigeria (CBN) oversees the Bank Verification Number (BVN) system.

Other agencies, including the Nigerian Communications Commission (NCC), Nigeria Immigration Service (NIS), Federal Inland Revenue Service (FIRS), Federal Road Safety Corps (FRSC), Corporate Affairs Commission (CAC), and the Independent National Electoral Commission (INEC), maintain separate databases for telecom subscribers, passports, taxes, driver’s licences, business registrations, and voter records.

Nigeria has spent nearly two decades trying to solve this problem. The National Identity Management Commission Act of 2007 established the National Identity Management System (NIMS) as a central identity framework designed to connect government databases.

The most aggressive push was in 2020 when the government mandated the linkage of SIM cards to National Identification Numbers (NIN), aiming to connect telecom data with verified identities. Yet despite these efforts, institutional rivalries and concerns over data ownership continue to keep many systems running in parallel rather than as part of a unified digital infrastructure.

The stakes have become even higher as Nigeria accelerates its AI ambitions. In 2025, the country launched a National Artificial Intelligence Strategy and unveiled N-Atlas, Africa’s first government-backed multilingual large language model. But experts say the challenge is no longer about vision; it is about execution.

“Today, we move from policy to progress,” said Kashifu Inuwa Abdullahi, Director-General of the National Information Technology Development Agency (NITDA), in remarks delivered by Emmanuel Edet, Acting Director of Regulation and Compliance at the AI Summit Nigeria in Abuja on Tuesday. “The true measure of success is not the number of policies we publish, but the impact these policies create on the lives of average Nigerians.”

Achieving that impact, however, may depend on breaking down data silos across the government.

“Artificial intelligence does not run on algorithms alone,” Abdullahi said. “It runs on energy, compute capacity, data, talent, infrastructure, and most of all, trust.”

The statement reflects a growing global understanding that AI leadership depends not only on technical capability but also on institutional readiness. Countries leading in AI adoption are not necessarily those building frontier models; they are often those that have succeeded in integrating data and digitising public services.

John Edokpolo, Microsoft’s Head of Government Affairs for Africa, pointed to countries such as the United Arab Emirates and Singapore as examples.

“These countries are not necessarily leading in chip design or model development,” he said. “What they have done well is digitise governance and create centralised systems that enable data sharing and AI diffusion.”

The problem is not technical

According to Edet, government agencies classify and manage data differently, creating inconsistencies that make information exchange difficult.

“We carried out a survey and realised that different government agencies classify data in different ways,” he explained. “How do you harmonise this so that once you have a class of data, you know what type of data you expect and how to manage that data across all agencies?”

Without standardised classifications, AI systems cannot effectively aggregate information from multiple sources. A healthcare AI platform, for instance, may struggle to combine hospital records if institutions use different formats or standards.

To address this, NITDA said it is working through the National Cloud Policy and developing frameworks for data exchange. Yet the process is proving more complex than anticipated.

“The challenges are enormous,” Edet admitted during a panel session where he represented himself. “It will take longer than we anticipated.”

Government agencies increasingly recognise that data carries value, according to Edet. In some cases, that value translates into institutional relevance, influence, or future monetisation opportunities.

“A lot of government agencies understand that data has value,” Edet said. “As far as they are concerned, sharing data is giving up that value.”

This creates a paradox. AI systems require integrated datasets to generate insights, yet the very institutions that hold these datasets are often reluctant to share them.

“Nobody wants to be irrelevant in any system,” he added.

The result is a fragmented digital ecosystem where valuable information remains locked within institutional boundaries.

Trust is the currency of AI

If data is the fuel of AI, trust may be its currency. AI deployment cannot succeed without public confidence in how data is collected, processed and used.

Nigeria’s regulators appear aware of the risks.

Babatunde Bamigboye, Head, Legal Enforcement and Regulations Department, at the Nigeria Data Protection Commission (NDPC) highlighted the importance of lawful, fair and transparent data processing under the Nigeria Data Protection Act.

The rise of AI introduces new challenges because AI systems often require vast amounts of data to function effectively.

“Collecting a million data points may be permissible,” Bamigboye explained, “but the question is whether the purpose is legitimate in relation to the data subject.”

This becomes especially important when dealing with vulnerable populations such as children.

Using AI to provide educational tools for underserved communities may align with public interest. Using similar systems to manipulate consumer behaviour, however, may not.

The principle, regulators say, is simple: AI innovation must remain human-centred.