MTN Group is concentrating on R30-billion in “worth creation” from AI over the subsequent three to 5 years, with roughly half anticipated to return from utilizing AI to run the enterprise extra cheaply and the steadiness cut up between client and enterprise functions.
Setting out the plan on the group’s capital markets day on Wednesday, chief know-how and knowledge officer Charles Molapisi mentioned MTN would prioritise inside effectivity – the guess it’s most assured about – whereas transferring extra cautiously on the riskier, capital-intensive push to promote AI infrastructure to different companies.
AI at MTN will not be one programme however a portfolio of three, every with a special threat profile and funding mannequin:
- AI Inside makes use of AI to run MTN itself – networks, power, finance, HR, authorized and fraud.
- AI for B2C makes use of AI to promote extra to shoppers via hyper-personalised gives.
- AI for B2B is the industrial play: renting out computing energy, GPUs and edge capability.
Learn: MTN South Africa hunts as much as R6-billion in financial savings
Every is measured in another way. AI Inside is a margin lever – worth is solely working prices saved plus capital spending prevented. AI for B2C is gated by buyer adoption and the price to serve. AI for B2B is the largest potential prize however the riskiest – capital-intensive and contestable, hinging on power prices and “token economics”, the worth of manufacturing AI output.
“We are going to prioritise AI Inside, however we’re executing B2C and B2B – we’re just a bit extra nuanced by way of timing and capital allocation,” Molapisi mentioned.
The group goes after its price base first. The logic for that is in MTN’s revenue assertion: community and IT account for round 55% of group working expenditure and roughly 80% of capital expenditure. “It due to this fact follows that that is what we’ve to go after if we need to create worth,” he mentioned. The most important single goal is power.
A number of inside use instances are already dwell:
- Power optimisation manages energy draw at base stations, with early deployments within the Western Cape; Molapisi argued the distinction from previous telco power guarantees is scale, with AI in a position to interrogate vastly extra parameters.
- Autonomous networks use brokers to triage incidents in South Africa and, through an approval course of, tune some parameters mechanically.
- Fibre-cut sensing turns the fibre itself right into a sensor that listens for micro-vibrations and flags digging earlier than the spade goes in – aiming to break down restore home windows that may simply run to eight hours or extra.
- Sensible capex forecasts the visitors and income profile of a locality six to 12 months out, so MTN can web site cellular (AI RAN, or radio entry community) and stuck (AI FAN, or fastened entry community) builds with way more precision.
- Within the again workplace, authorized contracts are being loaded right into a single vault so AI can draft, amend and flag regulatory adjustments.
Promoting to shoppers
On the buyer facet, the flagship is NBx 2.0, a “next-best-action” engine that targets people utilizing richer information. It’s dwell throughout six markets – South Africa, Nigeria, Ghana, Cameroon, Zambia and Uganda – reaching 44 million clients, and is scaling to all. Alongside it sit Telco GPT, an assistant for patrons and name centre brokers, and Zigi, a conversational software that now works by textual content and voice in native languages.
Learn: MTN to show its African towers into an AI inference grid
The use case Molapisi returned to most was Sim registration in Nigeria. Till just lately, greater than 200 employees manually “eyeballed” incoming data, matching biometric information towards the central database. MTN has changed that with 13 AI brokers utilizing laptop imaginative and prescient, now dealing with the majority of processing autonomously and, it says, sooner and extra precisely. It’s scaling to Cameroon, Côte d’Ivoire, Eswatini, Ghana and Zambia – with governance and “human within the loop” safeguards. A fourth dwell case, income assurance, scans 10 000-plus information factors throughout billions of day by day transactions to catch anomalies “people can’t diagnose”.
To carry itself to account, MTN has constructed an “AI depth system” with a 2028 goal: in its prime six markets, 80% of AI’s bottom-line impression ought to come from AI Inside. That rests on 70% of the workforce skilled, 40% “mannequin and agentic penetration”, and 80% maturity in underlying know-how and processes.
Entry will probably be focused to roles the place worth justifies price.
For now, the surest returns are those MTN can wring from its personal operations. – © 2026 NewsCentral Media
