New GSMA report shows how Africa’s top mobile operators are building AI language models in African languages

GSMA published its Mobile Economy Africa 2026 report on Tuesday, and one section details how six of Africa’s largest mobile network operators have partnered with the industry body to build artificial intelligence (AI) language models trained on African languages. The effort is aimed at closing a gap that leaves most of the continent unable to use mainstream AI tools in the languages people actually speak.

Known as the G6, the operators, Airtel, Axian Telecom, Ethio Telecom, MTN, Orange, and Vodacom, first unveiled the initiative at Mobile World Congress (MWC) Kigali in October 2025, under the banner “AI language models in Africa, by Africa, for Africa.”

The collaboration brings together operators, researchers, startups, and civil society groups to tackle four barriers standing in the way of African-led AI development: data, compute, talent, and policy.

The stakes are significant. Africa is home to more than 2,000 languages, accounting for over 30% of the world’s total, yet large language models (LLMs) are trained mostly on English and other high-resource languages. That leaves hundreds of millions of people unable to use AI tools by voice or text in their own languages.

A working Swahili model

The initiative has already produced a tangible result. At MWC Barcelona 2026, developers demonstrated the first open Swahili reasoning model, built in partnership with MeetKai Zambia. The model can browse and translate online content in Swahili, offering a template that could be replicated for other African languages.

Closing this language gap is not only a social good, the report argues, it carries real economic weight. Operators that deploy AI effectively expand the addressable market for digital services to the hundreds of millions of Africans currently excluded by language barriers, helping push mobile’s contribution to Africa’s economy from $240 billion in 2025 to a projected $290 billion by 2030.

Building the data, compute, and talent base

Beyond the language models themselves, the report points to several pieces of supporting infrastructure falling into place. Cassava Technologies has opened an AI factory in South Africa, renting out GPU computing power and AI services to African developers, businesses, and mobile operators, while running a network-management system built on Nvidia’s Blueprints framework to automate performance tuning across Africa’s patchwork of network equipment.

On skills, the MTN Skills Academy, a zero-rated platform open across 16 African markets, is training the workforce AI rollouts will need. On data, Google’s WAXAL project gives that workforce something to build with. Released in February 2026 after three years of work with African universities, including Makerere University, the University of Ghana, and Addis Ababa University, the open dataset captures speech across 27 Sub-Saharan African languages spoken by more than 100 million people, designed to train voice AI and speech-recognition tools.

Separately, the GSMA teamed up with Zindi, an African data science platform, to launch the African Trust & Safety LLM Challenge in March 2026. The result is a reusable safety benchmark that tests how AI models perform across more than 2,000 of the continent’s languages.

Victoria Fakiya – Senior Writer

Techpoint Digest

Stop struggling to find your tech career path

Discover in-demand tech skills and build a standout portfolio in this FREE 5-day email course

Operators are backing the infrastructure too

The push extends to hard infrastructure. In April 2026, MTN took part in a $45 million funding round for ODC, an AI telecoms startup building network technology designed for Africa’s mix of equipment vendors and patchy power supply. The deal signals operators moving beyond simply renting capacity from global cloud providers, toward investing directly in the technology layer that will shape network performance and costs over the next decade.

The G6 initiative isn’t the only effort chipping away at Africa’s AI language gap. Orange, one of the G6 members, had already partnered with OpenAI and Meta in 2024 to build AI models for African languages, beginning with Wolof and Pulaar before expanding toward Swahili and other languages. In July 2025, Google also invested in AI tools supporting more than 40 African languages, and the Gates Foundation has given Masakhane, a research collective, a $5 million grant to develop African language models, widening the coalition working to close the gap.