Vodacom's group chief information officer explains why the group runs its AI platforms in parallel – cloud-based and fully on-premise – to survive Africa's 40-plus regulatory regimes.
Vodacom group CIO Mohamed Sami says the telecoms giant builds its AI platforms twice. Once to run in the cloud, and once to run entirely on-premise, with no cloud dependency at all.
It's an unusual, resource-heavy approach, but it's non-negotiable for the Egyptian-born CIO.
“We had to build our platforms and solutions in a way that allows them to work either with public cloud-friendly markets or completely in an on-premise environment,” explains Mohamed in an exclusive sit-down with CIO South Africa.
“This was not easy, but we were leading the way in building on-premise capabilities and security by design to protect our customers’ data.”
Two extremes, one company
That duplication exists because Vodacom group operates in multiple regulatory environments.
“At Vodacom, we operate across varying regulatory landscapes – from public cloud-friendly markets like South Africa and Kenya, to stricter regulatory environments like Ethiopia and Egypt,” Mohamed says.
“Africa is not one market. Infrastructure, language, regulations, device capability, affordability and digital maturity vary significantly,” says Mohamed. “Transformation must take this local context into consideration so that it is resilient, inclusive, affordable and locally relevant.”
Security first, then scale
Underneath both versions of the platform sits the same foundation: a federated data layer Vodacom built to replace years of fragmented data lakes and legacy warehouses, plus what Mohamed calls “data products” – correlated data assets built around specific business needs, from network performance to customer care.
On top of that sits AI Booster, Vodacom group's internal security-by-design layer. “The objective is to guarantee the trust and sovereignty of our customers’ data through a platform concept where users can directly utilise AI while security happens in the background,” Mohamed says.
That same discipline extends to how Vodacom governs AI agents and manages what it spends on them. “We are also building AI agents, which require proper harnessing of the outcome to control hallucination and enable autonomy, as well as pan-African platforms to control tokenisation and AI economics,” he adds.
Mohamed is equally clear that responsible use doesn't stop at infrastructure. “Customers should know when they are interacting with AI, and for high-impact decisions, we may preserve a meaningful route to final human intervention,” he says.
Ask Mohamed what “winning” at AI in Africa actually looks like, and he doesn't point to model counts or platform sophistication. “In Africa, the best digital solution is not the one with the most features; it is the one that works reliably, affordably and inclusively in the customer's real environment,” Mohamed concludes.
Mohamed believes Africa has a unique opportunity in the age of AI, not just to consume technology but to build it. He is passionate about developing AI talent across the continent and sees immense potential for African innovators to create solutions tailored to local challenges, from financial inclusion and healthcare to agriculture and education. “The future of AI in Africa will be shaped by Africans solving African problems,” he says.
