Nigeria races toward AI data centre ambitions amid energy gaps

Nigeria is pushing to bring fully artificial intelligence-ready data centres online in 2026, betting that local compute capacity will anchor its digital economy, but industry executives warn that unreliable power and a shortage of skilled workers could undermine those ambitions.

Construction is underway across Nigeria, where telecom operators and global infrastructure firms are committing close to $1 billion to next-generation facilities designed to handle high-density graphics processing units (GPUs) and advanced cooling systems required for AI workloads.

At the centre of the expansion is the Sifiso Dabengwa Data Centre in Ikeja, developed by MTN Group under its Genova infrastructure unit. The first phase of the 4.5-megawatt, Tier III-compliant facility began operations in July 2025.

A second phase, due in the second half of 2026, will add AI-optimised GPU infrastructure at a cost estimated between $240 million and $250 million.

Read also: FG’s planned Data Centre signals shift in approach to housing governance

Ralph Mupita, chief executive, MTN Group said the company is in talks with U.S. and European partners to co-fund and expand AI-focused data centres across Africa, part of a strategy to diversify beyond traditional telecom services.

The move reflects a broader shift among African operators seeking to capture more value from cloud computing and AI, rather than relying on overseas hosting.

Global interconnection firm Equinix plans to open its $22 million LG3 facility on Victoria Island in the first quarter of 2026, marking its first newly built site in West Africa. Abayomi Adebanjo, director of legal at Equinix West Africa, said regulatory stability and infrastructure tailored to local realities would be critical to success.

Airtel Nigeria, through its Nxtra platform, is investing $120 million in a hyperscale facility at Eko Atlantic aimed at high-performance and AI workloads. Johnson Agogbua, chief executive of Kasi Cloud, has outlined plans for a 100-megawatt AI campus in Lekki, with 5.5 megawatts expected online in the second quarter of 2026.

Despite the momentum, executives say electricity remains the industry’s biggest constraint.

Speaking at Hyperscalers Convergence Africa 2025 in Lagos, Bill Kleyman, chief executive of Apolo.us and executive chair for Data Center Programs at Informa, said data-centre power demand in Africa is rising by 20 percent to 25 percent annually and could reach 8,000 gigawatt-hours.

“Success requires two things: power and bravery,” Kleyman said, warning that rapid AI adoption is driving rack densities far beyond what many facilities were originally designed to handle.

Wilson Eigbadon, regional account manager at Vertiv, said traditional grid supply in many African markets is too unstable for AI-era infrastructure.

“Africa is heading into an era where data centres will have to bring their own power,” he said, arguing that operators will need hybrid systems combining gas, solar and battery storage to guarantee uptime.

Ayotunde Coker, chief executive of Open Access Data Centres, said even advanced economies are exploring small modular nuclear reactors to support hyperscale AI facilities, underscoring how central energy security has become to the global AI race.

Bukola Ajayi, general manager of architecture and enterprise IT at MTN Nigeria, said reliable electricity and connectivity are non-negotiable for AI readiness. High-density racks and advanced cooling systems cannot operate consistently on unstable grids, she said.

The challenge is compounded by broader infrastructure gaps. Around 600 million Africans lack reliable electricity access, even as new data centres consume power equivalent to small towns. Operators say diesel backup can account for up to 40 percent of operating costs, squeezing margins and raising environmental concerns.

Beyond energy, policymakers frame the data-centre buildout as a sovereignty issue.

Temitope Osunrinde, executive director of Africa Hyperscalers, said Africa accounts for about 18 percent of the world’s population but less than two percent of global data-centre capacity and under one percent of compute power. Roughly 80 percent of African data is hosted offshore.

Industry estimates suggest Nigeria spends about $850 million annually on foreign cloud services, contributing to foreign exchange pressures.

Muhammed Rudman, chief executive of the Internet Exchange Point of Nigeria, said efforts are underway to retain more traffic domestically and reduce reliance on international routes.

Read also: ‘Why data centres will need their own power’

Kashifu Inuwa Abdullahi, director general of the National Information Technology Development Agency, has described AI as a greenfield opportunity for Nigeria to address challenges in areas such as agriculture and health, citing regulatory reforms and a 3 Million Technical Talent programme aimed at expanding the country’s skills base.

Still, analysts say sovereignty ambitions must be commercially viable. Guy Zibi, managing partner at Xalam Analytics, has argued that local compute investments will only be sustainable if backed by strong demand and balanced partnerships.

Talent retention poses another risk. Agogbua said that while power and connectivity may eventually be addressed, training and keeping specialised engineers could become the deeper bottleneck.

Roger Shutte, general manager for infrastructure and cloud engineering at MTN Nigeria, said Nigeria must ensure newly trained professionals remain to support domestic enterprises.

Other industry leaders, including executives from Africa Finance Corporation, AFIDA, Nokia and IX Africa Data Centres, have called for closer coordination between energy policy, telecom infrastructure and investment frameworks.

Market forecasts show Nigeria’s data-centre sector expanding from just over $300 million in 2025 to nearly $800 million by 2031, supported by new subsea cables such as Equiano and 2Africa and rising cloud adoption.

As 2026 approaches, Nigeria’s AI infrastructure is taking shape. Whether the new facilities become engines of digital growth or expensive underutilised assets may depend less on construction timelines than on the country’s ability to secure reliable power, retain skilled workers and sustain demand in an increasingly competitive global AI market.