Edition of Wednesday, November 11, 2026
Business Africa

Vodacom’s AI Data Centre Expansion Will Depend on Local Partners

Vodacom says it is looking for partners as it explores AI data-centre opportunities in South Africa and Kenya. Egypt offers an early test of the partnership-led model, but expansion will depend on demand, power and project economics.

By 6 min read
vodacom-africa-ai-data-centres-partners-egypt-kenya-south-africa-2026
Photo: Eukoha/Wikimedia Commons — CC BY-SA 4.0.

Vodacom’s next infrastructure bet in Africa may be built with partners rather than by the telecom group alone. Chief executive Shameel Joosub said the company is seeking partners for artificial-intelligence data-centre opportunities, with South Africa and Kenya among the markets under consideration after work in Egypt. The distinction matters: this is an exploration of where and how to expand, not a published commitment to build a facility in every named country.

The strategy reflects a practical tension in the AI economy. Demand for computing capacity is rising, but data centres require substantial long-term capital and dependable electricity, network links and customers. For a regional operator, a partner can contribute specialist expertise, infrastructure or financing while Vodacom brings connectivity, enterprise relationships and experience operating across markets. The model only works, however, if responsibilities and commercial returns are clear.

Egypt is a reference point, not a template by default

Vodacom’s Egypt activity provides a concrete starting point for the discussion. In September, the group announced collaborations involving Vodafone Business, Cassava Technologies and Elsewedy Electric around data-centre infrastructure in Egypt. The company also described a separate collaboration with Cassava focused on sovereign AI infrastructure, including NVIDIA computing technology. These are related parts of an expanding digital-infrastructure effort, but they should not be collapsed into one project: a broader data-centre development and infrastructure dedicated to AI applications have distinct scopes.

Those announcements show the kinds of capabilities a partnership can assemble: local industrial and infrastructure experience, data-centre development, telecom reach and specialist computing. They do not establish that the same partners, configuration, capacity or timetable will be used in South Africa or Kenya. Each market has its own customers, power arrangements, regulation and competitive landscape. A repeatable regional strategy will need local adaptation rather than a simple copy of the Egypt model.

Joosub’s comments, reported following a Bloomberg interview, put South Africa and Kenya among the markets being considered alongside Egypt. The group’s official interim results, released on 10 November, offer useful corporate context: Vodacom reported 250 million customers and R102.5 billion in revenue for the half-year, with continued investment across its markets. Those figures indicate the scale of the wider operator, not a specific budget allocation for new AI data centres. No public project-level investment, construction timetable or partner terms for the potential South African and Kenyan sites were disclosed in the cited reporting.

Partnerships can share risk, but cannot manufacture demand

A partner-led approach can reduce the amount of expertise Vodacom must assemble on its own and distribute capital exposure. It may also help align a facility with local requirements, especially when a partner already understands land, power procurement, permitting or enterprise demand. For a telecom operator, proximity between data centres and strong fibre or mobile networks may create opportunities to sell secure hosting, cloud connectivity and AI-related services to businesses and public institutions.

Yet a data centre is not commercially sound merely because AI is attracting attention. Investors need credible evidence that customers will use capacity and pay for it over time. That can come from contracted workloads, enterprise demand, cloud-provider commitments or a phased plan that brings capacity online as utilization grows. Without enough committed demand, a large facility risks carrying expensive idle capacity. Without clarity on who owns the hardware, bears utilization risk and controls customer relationships, a partnership can simply make accountability harder to see.

The most persuasive expansion case would therefore connect each proposed site to an identifiable customer base and a staged investment plan. South Africa and Kenya are not interchangeable markets, and a country’s status as a major regional economy does not itself prove that a new facility will achieve attractive utilization. Local businesses, banks, telecom customers, public agencies and cloud providers may have different requirements around latency, security, data location and service levels. The operator and its partners would need to show which of those needs are addressable, and at what cost.

Power, connectivity and trust are part of the product

Reliable electricity is a core operating requirement, not a background detail. Data-centre customers expect continuous service, so developers must plan for grid conditions, backup systems, energy costs and resilience. Connectivity matters just as much: a facility needs dependable routes to users and other networks, with adequate redundancy. These choices affect operating costs and whether customers can rely on the service for workloads that cannot tolerate interruption.

There is also a governance question. “Sovereign AI” and local hosting can appeal to customers seeking greater control over data and computing, but the phrase is meaningful only when the service’s ownership, data handling, security responsibilities and jurisdiction are transparent. Businesses will want to know who can access information, where it is processed, how incidents are managed and whether the infrastructure meets their compliance obligations. Trust is built through verifiable controls and service performance, not a label alone.

Vodacom’s advantage may be its ability to connect infrastructure to an established base of telecom and enterprise services. That could make the company a useful commercial anchor or distribution partner, especially if it can bundle connectivity, cloud access and support. But the operator will still need specialist data-centre capabilities and credible counterparties. The strategic test is not simply whether partners can announce a project; it is whether they can deliver uptime, security and cost-effective capacity that customers actually adopt.

The next proof point is project detail

For now, the news is best read as a directional signal: Vodacom sees AI infrastructure as a potential extension of its African digital business and prefers to explore expansion with partners. The next evidence investors and customers should look for is concrete: named counterparties for each market, a defined facility scope, financing and ownership arrangements, power and connectivity plans, customer commitments, and phased capacity targets. These details would distinguish a serious project pipeline from broad strategic intent.

There is a sound logic to exploring several markets before committing capital. Egypt provides an active partnership case; South Africa and Kenya are being considered as possible next locations. Whether the opportunity becomes a regional platform will depend on disciplined sequencing. If partners can match local demand with reliable operations and transparent economics, Vodacom’s network could help turn computing capacity into a service businesses can use. If those fundamentals remain unclear, the prudent outcome may be to stage, resize or defer projects rather than build ahead of demand.

That is why partnership is more than a way to divide a bill. It is the mechanism through which the company must coordinate capital, technology, power, regulation and customers across distinct African markets. The ambition is notable; the quality of the eventual project agreements will determine whether it becomes durable infrastructure or remains an exploratory plan.

Sources

Africa technologyartificial intelligencedata centresEgyptKenyaSouth AfricaVodacom

Privacy Policy