MTN Group is targeting 150MW of capacity in a first phase spanning South Africa and Nigeria, group CEO Ralph Mupita told a media roundtable on Tuesday. The group will hold only a minority stake in the UAE-backed data-centre venture behind it, he said.

The comments put hard numbers to a deal MTN described only in outline in its results for the six months ended 30 June 2026.

The results booklet confirmed that, after the reporting period, MTN Digital Infrastructure had “entered into a strategic partnership with a UAE-backed data centre investment platform to accelerate the development of AI-ready digital infrastructure platforms across Africa”.

The vehicle, named Africa Data Hub Holding, “will serve as the platform through which future digital infrastructure opportunities will be developed and scaled across key African markets”.

What the booklet did not disclose was MTN’s stake, the capacity planned or the capital committed. Nor did MTN name the partner. Mupita filled in some of those blanks on Tuesday.

“For data centres, our approach is to partner with third parties with the necessary skills and capabilities, like the mentioned UAE-backed deal. We will be a minority investor,” he said. “For now, only in South Africa and Nigeria, we are in that partnership, looking at 150MW as phase 1. We will build out as demand requires. We are making sure we acquire the land, and necessary power arrangements, for multiple phases to build out these data centres as needed over time.”

Enterprise and cloud

The minority-investor stance is consistent with the capital-light approach MTN has signalled as it lines up partners for an African AI data-centre play rather than funding hyperscale capacity off its own balance sheet. It also fits the repositioning under the group’s Ambition 2030 strategy, which leans on partnerships in fintech and digital infrastructure while the group keeps a tight rein on leverage. MTN closed the period with net debt to Ebitda of 0.3x and did not declare an interim dividend, though it confirmed that a R6-billion share buyback would start.

The demand behind the data centre push is enterprise and cloud, not the retail mobile market. MTN says the platform will address “growing demand for cloud, enterprise and data-intensive services” across Africa, and it expects the continent’s enterprise market to grow around 1.6 times by 2030.

Read: MTN’s Iranian dead end

Those are different customers and workloads from the connectivity business that still generates most of its money, where group service revenue climbed 17.5% in constant currency to R115.3-billion in the half. MTN had earlier flagged plans to turn its African tower estate into an AI inference grid.

Asked which AI use cases excite him most, Mupita reached not for customer-facing products but for the network itself.

“With AI, optimising energy efficiency across our networks is a no-brainer. We’re looking at [Cape Town] already as a test base, to dynamically bring power consumption down,” he said. “So, it might sound boring, but to me network-related efficiency is the most interesting at the moment.”

Energy is one of MTN’s largest operating costs, and the group has said its first AI target is itself, applying the technology to internal operations before selling it on.

Routing between models

On frontier models, Mupita said MTN was less interested in backing a single winner than in building a system that can route work to whichever model performs best.

“AI frontier models, and how we integrate them, is an ongoing question – especially for enterprise. It is a fast-evolving space that keeps surprising us every two or three months,” he said. “We are looking at, how do we create an open architected system that integrates different types of models, like frontier models combined with open-weighted models from China, to automatically find the most optimal model for the task at hand. That is one approach we are looking at.”

Mupita also sits on a new Global AI Commission, taking forward the work of the Broadband Commission. He was named to the commission earlier this year.

MTN plans to use AI to optimise energy efficiency in its networks

“It was an interesting discussion, including on how to make sure the AI race doesn’t create a digital underclass,” he said. “It is a geopolitical issue as well, to make sure the Global South has access. And that AI companies and role players operate ethically.”