Coker sat down with Capacity Media to share candid insights into the shifting landscape of the continent’s infrastructure, investment climate, and the growing impact of AI.
In a year marked by rising digital demand, increased data sovereignty ambitions, and a surge in AI innovation, Coker believes Africa is not only catching up, it’s preparing to lead.
“Everything is changing rapidly, particularly over the last 12 to 18 months,” says Coker. “And AI is right at the centre of it.”
While AI has been evolving for decades, the advent of generative AI and agentic AI has fast-tracked changes in compute demands, data centre design, and network interconnectivity.
“We’ve moved beyond the era of big data. Now, we’re dealing with what I call gargantuan data,” Coker explains.
The compute requirements for training and deploying advanced AI models, such as large language models (LLMs), are staggering. “10 years ago, a 2MW data centre was considered substantial. Now we’re talking about 24MW, even 36MW on the same footprint. Some operators are already planning for gigawatt-level infrastructure.”
The catalyst? A combination of cheaper, more powerful GPUs and an evolving AI architecture that requires real-time or near real-time compute capability at the network edge. “Agentic AI brings a new demand: low latency, high density, and immediate responsiveness,” says Coker. “It’s machine-to-machine and machine-to-human, and it’s changing how we plan infrastructure from the ground up.”
Crucially, these developments are not being imported wholesale from the global north. Africa is developing its own capabilities, and its own narrative.
“There’s a lot of innovation happening locally,” Coker points out. “We’re seeing AI Centres of Excellence emerge in Nigeria, South Africa, Kenya. Many of the AI coding teams powering global platforms include African talent working remotely.”
This trend reflects a broader shift toward sovereign AI; AI solutions that are developed, hosted, and deployed within African jurisdictions to meet local policy, language, and regulatory requirements.
The rise of sovereign AI also reinforces the importance of data sovereignty, localisation, and secure infrastructure: three forces shaping investment decisions across the continent.
The shift to AI-ready infrastructure is forcing data centre operators and investors to reconsider traditional models. “You can’t just pick up the next design template and say it’s AI-ready,” Coker warns. “It’s not just about putting high-performance servers in racks.”
AI workloads demand high-density compute with significant power draw, often exceeding 100kW per rack. This introduces new engineering challenges: heat dissipation, latency reduction, and power distribution, to name a few.
“What’s emerging is a hybrid model,” he explains. “Core liquid cooling combined with traditional air cooling, high-density zones alongside general compute zones — all underpinned by a flexible, modular shell-and-core design.”
Connectivity is also paramount. “The volume of interconnect needed has gone through the roof. We’re no longer talking about 100G or 400G links, we’re talking about petabits,” says Coker. Internet Exchanges (IXs), he adds, are set to scale massively to cope with this surge. “They’re going to become mega-IMCs.”
10000740
So where should investment focus?
Coker is clear: “We need to build the right infrastructure: hyperscale, high-density, high-connectivity, and we need to do it quickly.
Using the analogy of hotels, he explains: “Boutique hotels have their place. But if you want to attract global players, you need five-star, full-service properties. It’s the same with data centres. Lagos and Johannesburg are our hyperscale hubs. Nairobi is emerging. We need more of them.”
However, Africa’s vast geography demands decentralised growth. A six-hour flight from Lagos gets you to Nairobi or Dakar, distances comparable to flying from Geneva to Dubai. “You can’t serve the continent from three hubs,” Coker says. “We need a tiered model: mega hubs in key metros, and regional edge facilities that bring content closer to users.”
Yet regulation remains a hurdle. Country-by-country data protection laws can fragment progress. “We need an ‘Africa DPR’ — a harmonised data protection regulation, like the EU’s GDPR,” he says. The Malabo Convention, long-standing but underutilised framework for African data protection, dormant for nearly a decade, could be revived as a unifying mechanism.
Of course, none of this is possible without reliable, scalable power. Coker is realistic about the challenge. “You can’t run a 24MW data centre without thinking about where 45MW of power is coming from and that’s just for IT load and redundancy.”
In some regions, that means captive generation using natural gas or hydro. Elsewhere, it means partnerships with municipal utilities and investment in grid improvements. “Gas, despite being a fossil fuel, is still the cleanest transitional option for many African markets,” he adds.
Sustainability is on the agenda too, but with pragmatism. “It’s not about chasing perfection. It’s about designing systems that work for each region, hydro in the DRC, geothermal in Kenya, solar where viable and reducing diesel reliance wherever possible.”
One of the starkest challenges facing Africa’s AI ambitions is access to GPU infrastructure. “Africa has largely been bypassed in the global GPU race,” Coker says. “We need localised GPU-as-a-Service models to catch up and we’re starting to see interest.”
He points to recent efforts by chip manufacturers and alternative providers looking to expand beyond the Nvidia-dominated ecosystem. “It may not be at the same performance level yet, but diversification is happening, and Africa needs to be part of that supply chain.”
This shift also underlines the importance of cloud infrastructure. “AI has stolen the limelight, but it still needs cloud to run,” he reminds. “You can’t build AI systems without the underlying hyperscale cloud fabric and that has to be closer to the point of use.”
Coker is cautiously optimistic about the next five years. “We’re seeing bullish momentum in some markets. Cape Town, Durban, Lagos they’re building capacity fast. But it’s not fast enough. The demand is outpacing the supply curve.”
He calls for faster execution, stronger regulatory clarity, and more homegrown executive expertise. “Africa needs leaders who can deliver digital infrastructure at speed, with foresight. And we need to train them now; we can’t wait for the global market to catch up.”
As for market consolidation, Coker expects moderate M&A activity driven by the need for capital efficiency and operational scale. “You’ll see aggregation of infrastructure and expertise, but only where it adds strategic value. This market is still young. There’s room for collaboration before consolidation.”
If there’s one message Coker wants investors and policymakers to take away, it’s this: Africa is ready, but it needs infrastructure to match its ambition.
“We’ve got the talent. We’ve got the market. We’ve got the innovation. Now we need the hyperscale hotels of compute. And we need them now.”
The next two years, he says, are critical. “You can’t wait until you need the capacity to start building it. By then, it’s too late. Africa’s digital future depends on what we do right now.”
RELATED STORIES
Capacity Middle East 2026
Capacity Middle East is the region’s leading digital infrastructure event, uniting over 3,500 executives from more than 90 countries for visionary content and unrivalled networking and business opportunities.
