MTN Group is embarking on an ambitious initiative to transform its African tower network into a distributed AI computing grid. The plan, articulated by chief technology officer Charles Molapisi at a Johannesburg event, aims to install open GPU infrastructure at cellular base stations. This dual-use hardware will support both cellular network operations and edge AI inference tasks.
Currently, each cellular tower houses a baseband unit specifically designed to facilitate the radio access network. MTN intends to replace this single-purpose equipment with versatile GPU configurations, allowing the towers to process AI workloads locally. Molapisi emphasised that this change would significantly reduce latency, as AI tasks that were typically routed back to centralised data centres could be executed closer to the point of data collection. For instance, a child’s online gaming experience could improve markedly, as processing could occur right at the local tower rather than at a distant facility.
This edge computing strategy is complemented by MTN’s plans to enhance its centralised AI infrastructure. In March, the company announced the development of two new AI-enabled data centres, one in South Africa and another in Nigeria, further solidifying its AI capabilities across the continent.
Molapisi outlined a comprehensive AI strategy for MTN that encompasses everything from securing semiconductor technology and establishing data centres to developing cloud platforms and collaborating with partners on applications. Additionally, MTN is investing in terrestrial fibre networks across various African markets, even those where it does not hold a GSM license, to address what Molapisi termed Africa’s missing digital “rails.”
This initiative is part of MTN’s Ambition 2030 strategy, which revolves around connectivity, fintech, and digital infrastructure. The tower-to-inference project marks a significant step toward a vision MTN has been promoting for over a year and includes an investment in ORAN Development Company, an AI networking startup, alongside tech giants such as NVIDIA and Cisco. Molapisi emphasised the principle of “sovereign AI,” highlighting the need for African countries to host their AI computing resources rather than relying on foreign systems.
Ultimately, MTN aims to develop the largest edge AI distribution network in Africa, cautioning that the continent must avoid merely exporting raw data, as it has with natural resources, and instead foster local intelligence generation from its data assets.
