Every $1-billion invested globally in AI-dedicated data centres requires about $125-million in related energy-sector investment, with two-thirds of this amount needed for grid upgrades and one-third for generation capacity, notes credit insurance and risk specialist Coface South Africa.

Global concerns mirror South Africa’s issue of energy, land and water constraints. The US faces delays of up to five years for grid interconnection.

In South Africa, the structural electricity constraints underscore the need for strategic planning and investment to sustain growth in the digital economy.

South Africa’s grid instability, transmission backlogs and persistent load energy shortages create even more acute challenges for energy-intensive data centre operations Coface says in its 'Data centres in the AI Age: stakes, limits and risks of a trillion-dollar gamble' report.

Globally, the acceleration of AI continues to reshape investment flows, digital infrastructure and economic growth.

According to the report, data centres, which are the backbone of AI development, are expanding at an unprecedented pace, driven by massive capital expenditure and a high concentration of activity in the US.

Coface's global analysis indicates that $475-billion will be invested in IT equipment for data centres this year.

However, $750-billion worth of ongoing or planned projects globally could be delayed by 2030 owing to energy infrastructure saturation, primarily in the US, but with knock-on effects on supply chains worldwide, it notes.

Additionally, the analysis found that about one-fifth of US GDP growth in the second quarter of this year was attributable to the surge in AI-dedicated data centres.

However, this exponential growth is encountering structural limitations, new vulnerabilities and potential imbalances, which have major implications for emerging digital economies, such as South Africa, the report notes.

South Africa, which is a data centre hub in Africa with multiple hyperscale expansions, faces energy constraints, water scarcity, rising operational costs and talent shortages.

These factors position the country at the intersection of global opportunity and domestic fragility, and determine the global environment in which South Africa’s digital economy must operate, as they affect everything from local project costs to equipment availability and investment appetite.

Further, AI has shifted from specialist labs into everyday use and this has triggered a sustained explosion in demand for digital infrastructure.

This trend is shaping South Africa’s digital landscape. The country is seeing global cloud providers expanding their local footprints and banks, insurers, retailers and mining houses increasingly adopting AI tools.

However, telecommunications and infrastructure operators are grappling with rising power and cooling demands, and escalating global equipment costs are driving up local data-centre and cloud-services pricing, the report highlights.

“The rapid expansion of data centre and AI infrastructure is colliding with technical, geographical and human limitations, and South Africa is no exception,” it emphasises.

Further, the shortage of skilled labour compounds these difficulties. Globally and locally, demand for specialised engineers, technicians and AI-infrastructure experts far outstrips supply, which is driving up costs and causing project delays, it adds.

South Africa stands to benefit from the global AI acceleration, but its domestic constraints mean the country must manage the associated risks proactively, Coface advises.

Meanwhile, the global data centre boom faces downside risk.

Among several emerging threats, uncertain demand forecasts is such a risk. Projections for global AI processing requirements vary by up to 80%, which raises the risk of overcapacity, the report notes.

Another downside risk is the knock-on effects across the value chain. Any global overcapacity shock would first affect co-location operators and then cascade across the entire supply chain impacting cloud providers, equipment manufacturers, installers and service partners.

South Africa, which is heavily dependent on imported IT hardware, would be exposed to price fluctuations and supply disruptions, Coface points out.

Additionally, revenue pressure and tighter margins also present a risk. Overexpansion could affect profitability across the sector, including for South African operators building data centres based on optimistic AI-adoption projections, the company says.

“The rise of data centres in the age of AI is a colossal gamble. It promises major advances but also exposes the economy to unprecedented risks of overheating and imbalance,” says Coface information and communications technology economist Aurélien Duthoit.

“For this boom to have a lasting impact, it must translate into real productivity gains and large-scale services beyond investment growth. At this stage, AI is fuelling economic expansion without profoundly transforming it,” he says.