.
Africa’s push to build a competitive artificial intelligence economy is entering a costly new phase, as technology companies and infrastructure providers move beyond adopting AI tools to investing in the computing, energy and regulatory systems required to run them locally.
The continent’s AI ambitions are increasingly being tested by the high cost of GPUs, data centres, reliable electricity, fibre networks, cooling systems and compliance infrastructure.
The shift reflects a growing realisation among African technology companies that controlling AI applications alone may not be enough if the computing power, data infrastructure and energy required to operate those systems remain outside the continent.
AI Infrastructure Race Accelerates
One of the most significant developments is the strategy by Zimbabwe-founded technology group Cassava Technologies, which has partnered with NVIDIA on plans to develop AI factories across Africa.
The initiative involves an investment programme of up to $720 million, with plans to deploy approximately 12,000 NVIDIA GPUs across multiple African markets.
The move represents a transition from simply consuming AI services provided by global cloud companies towards building the underlying infrastructure needed to process AI workloads closer to where African data is generated.
Nigeria is also emerging as a major market in this infrastructure race. The second phase of a major data-centre development in the country could require between $240 million and $250 million for AI-optimised GPU infrastructure.
Such projects demonstrate that establishing AI capacity involves far more than purchasing servers. Operators must also finance land, buildings, cooling systems, networking, fibre connectivity, security, backup systems and power infrastructure.
Power Emerges as AI’s Hidden Cost
For Africa, electricity could prove to be an even bigger constraint than access to GPUs.
The International Energy Agency estimates that Africa had the world’s lowest data-centre electricity consumption per capita in 2024, at less than 1 kWh per person, although this is expected to approach 2 kWh by the end of the decade.
Meanwhile, data-centre electricity demand across Africa could increase from about 0.4GW to 2.2GW by 2030, according to McKinsey estimates cited in the report.
The challenge is particularly significant because reliable electricity remains limited across much of the continent. The International Monetary Fund estimates that roughly half of sub-Saharan Africa’s population still lacks dependable access to electricity.
For AI data centres, intermittent power is particularly problematic because high-performance computing workloads require consistent electricity, cooling and backup systems.
Consequently, access to affordable and reliable energy is increasingly becoming a competitive factor in determining where AI infrastructure investments are located.
Data Centres Become Strategic Assets
South Africa currently accounts for roughly 70 percent of Africa’s data-centre capacity, while Nigeria, Kenya, Egypt and Morocco are also attracting infrastructure investment.
Kenya’s geothermal resources offer an additional advantage as companies seek relatively cleaner and more dependable sources of electricity.
The planned data-centre investment involving Microsoft and G42 in Kenya, reported at about $1 billion, further illustrates the growing relationship between energy availability and AI infrastructure.
For African countries competing for investment, reliable electricity, fibre connectivity, suitable land and regulatory certainty could increasingly determine which markets emerge as AI infrastructure hubs.
Compliance Adds to AI Investment Bill
The cost of building AI infrastructure does not end with physical deployment.
AI systems increasingly process sensitive personal, financial, health, business and government information, creating significant obligations around data protection, cybersecurity, cross-border data transfers and sector-specific regulation.
By early 2026, 44 African countries had data-protection laws, with 38 reported to have functional data-protection authorities.
For technology companies operating across multiple markets, regulatory fragmentation can translate into additional expenses involving privacy assessments, cybersecurity controls, documentation, audits, staff training and ongoing monitoring.
The European Union’s AI Act also adds another layer for African companies serving European customers or handling European data, increasing the compliance requirements attached to international operations.
Ownership Comes With a Heavy Price Tag
Although owning AI infrastructure can reduce dependence on external cloud providers, it also requires substantial upfront capital and continuous reinvestment.
GPU technology evolves rapidly, meaning companies must regularly upgrade equipment to remain competitive. Data-centre projects also require long-term financing for construction, energy systems, hardware and maintenance.
This creates a difficult choice for African startups and enterprises: rent computing power and absorb potentially rising operating costs, or invest heavily in infrastructure and assume the risks associated with ownership.
Africa’s Infrastructure Opportunity
Despite the challenges, the infrastructure gap also represents a major economic opportunity.
Africa currently accounts for less than 1 percent of global data-centre capacity, despite being home to roughly 18 percent of the world’s population.
Closing that gap could generate demand for engineers, data-centre technicians, cybersecurity specialists, fibre operators, energy companies, cooling experts and software developers.
The continent does not necessarily need to manufacture GPUs to capture value from the AI economy. Ownership of data centres, cloud platforms, energy systems, networks and related services could allow African companies to control more of the infrastructure underpinning AI development.
The IMF estimates that AI could boost sub-Saharan Africa’s economy by around 4 percent over the next decade if electricity, internet connectivity and digital skills improve.
But achieving that potential will require substantial investment.
Africa’s AI opportunity is therefore no longer simply about building smarter applications. It is increasingly about who owns the computing power, energy, data infrastructure and regulatory systems that make those applications possible.
