Africa needs cheaper AI to reap economic gains, says Enzi’s Udezue

Enzi’s Oji Udezue Says Africa Needs Cheaper AI

Africa’s ability to benefit from artificial intelligence will depend on bringing down the cost of using it, Oji Udezue, president of Enzi, told Moonshot by TechCabal in Lagos.

Udezue said reducing the cost of inference, the process through which AI models generate responses and perform tasks, was essential to making the technology affordable for businesses and individuals across the continent.

“Our goal is to make AI cheaper for Africa and America,” he said. “The future of AI really goes through inference.”

His argument shifts attention from the race to build more powerful models to the economics of putting them to work. For businesses using AI to answer customer enquiries or support employees, the expense recurs each time the system performs a task. Those costs can determine whether a promising application becomes viable for everyday use.

Udezue highlighted the gap between the cost of using AI and what many Africans earn, warning that access to a tool means little if people cannot afford to use it consistently.

The distinction matters as companies seek to integrate AI into their operations. A service affordable for occasional queries may prove too expensive when used throughout the working day or offered to thousands of customers. Lower inference costs could make those applications more commercially practical.

Affordability is one part of a wider infrastructure challenge. The World Bank’s Digital Progress and Trends Report 2025: Strengthening AI Foundations identifies connectivity, computing resources, locally relevant data and digital skills as foundations for AI adoption. Its findings underline the importance of the systems and capabilities needed to put the technology to use.

For African developers, that also means ensuring models can work in local languages and respond to the circumstances of the people using them. A cheaper service will have limited value if its answers are poorly suited to customers’ needs.

Udezue emphasised practical demonstrations as a way to show what developers could achieve. His focus was on turning improvements in the technology into tools people could afford to use.

The economic test, in his account, is whether AI becomes cheap enough to support routine work across a much wider market. More capable models may expand what the technology can do; the cost of running them will help determine who can benefit.

Five months after fabricated academic references forced South Africa to withdraw its draft national artificial intelligence (AI) policy, one of the two officials suspended over the scandal has resigned. Disciplinary proceedings against the other communications department staffer are still under way.

Solly Malatsi, South Africa’s communications minister, says the government remains confident that Cabinet will finalise the replacement AI policy before the end of March 2027, with public consultation expected to begin in April.

But the unfinished disciplinary process leaves questions about accountability for a scandal that embarrassed a government seeking to position South Africa as a continental leader in AI governance.

Malatsi told TechCabal in an interview on Thursday that the department could not allow the incident to pass without consequences, acknowledging the damage caused by AI-generated material use that went undetected during the AI policy’s preparation.

“Having such a major stain on an important policy is something that cannot just be allowed to slide,” he said.

The minister confirmed that the department suspended two officials for failing to disclose AI use in preparing material that contributed to fabricated references appearing in the policy.

“The appropriate processes were followed, and two officials were suspended,” Malatsi said. “The ongoing processes now relate to the one official who remains an employee of the department.”

In May, TechCabal reported that South Africa withdrew its draft AI policy after fictitious academic references were discovered. The setback triggered renewed scrutiny over how generative AI is used in policymaking and exposed weaknesses in government oversight.

Malatsi said the department’s director-general acted after he instructed that those responsible be held accountable.

“It was clear that there had been non-disclosure around the use of AI, which contributed to the hallucinations that emerged in the policy,” he said.

The minister described the withdrawal as necessary despite the damage it caused to the government’s policymaking process.

At CollectiveXchange 2026, a national digital skills and jobs summit, in Johannesburg earlier on Thursday, Malatsi acknowledged that the process had experienced a “false start.” He stressed that the government was working to complete its internal processes before opening the replacement policy to public consultation.

“Yes, it is very late, and that is the consequence of having a false start,” he told delegates.

However, Malatsi rejected the suggestion in the interview with TechCabal that the government had introduced a new deadline for the AI policy. He said the January 2027 reference and the end of the 2026/27 financial year were not separate dates, while reiterating his confidence that the policy would be finalised by the end of March.

The timetable places responsibility on the independent AI review panel team led by Professor Benjamin Rosman of the Machine Intelligence and Neural Discovery Institute at the University of the Witwatersrand to restore confidence in the policy after the original draft’s credibility collapsed.

Malatsi said the panel had been working at pace and that he remained confident it would deliver a draft Cabinet could finalise. When the AI policy enters public consultation, he said, government departments would have made inputs, followed by engagement with civil society, residents, the private sector and other stakeholders.

The replacement policy will need to address more than the failures exposed by the citation scandal. Despite the fabricated citations, Malatsi said the withdrawn draft contained credible provisions, including protections for South Africans’ rights and dignity online.

“As unfortunate as the withdrawal of the policy was, it was necessary given the stain that it had,” he stated. “Much of the policy’s substance was credible, and everyone who has engaged with it has acknowledged that.”

The minister also pointed to the need for safeguards governing the ethical use of AI, including avenues for redress when the technology is used improperly.

“The important thing for us is to ensure that when we create a regime for the ethical use and ethical adoption of AI, where the use is not ethical, there’s got to be an avenue for recourse,” he said.

True scale demands moving beyond surface-level integrations to robust execution. We’ve filtered the noise out of Moonshot 2026, optimising the conference strictly for high-calibre connections between startup founders, global financial operators, enterprise leaders and individuals rewiring Africa’s technical frameworks. Get 20% off Early Bird tickets for a limited time.