RACA Journal I July 2026 www. refrigerationandaircon. co. za 19
DATA CENTRE FEATURE SERIES
WATER USE IN AI DATA CENTRES: IS THE CONCERN OVERSTATED?
By Eamonn Ryan
Insights from Wojtek Piorko, managing director for Africa at Vertiv.
As artificial intelligence accelerates the global expansion of data centres, concerns about the sector’ s environmental footprint – particularly water consumption – have gained prominence. However, according to Wojtek Piorko, the scale of water use in modern data centres is often misunderstood and, in some cases, overstated.
Speaking during a Q & A session at the seminar AI Unlocked: What‘ AI-ready’ means for 2026 decisions, Piorko addressed questions about cooling technologies, water management and the broader context of data infrastructure development in Africa. His message was that while water management is important, the narrative that data centres are excessively water-intensive does not always reflect the realities of modern cooling technologies.
CLOSED-LOOP COOLING REDUCES WATER DEMAND Historically, many large data centres relied on cooling towers to dissipate heat generated by servers. These systems can consume significant quantities of water, particularly in hot climates. As a result, concerns have emerged about the impact of data centres on already strained water resources.
However, Piorko explained that current-generation facilities increasingly rely on closed-loop cooling systems, which dramatically reduce water consumption. In these systems, water circulates continuously within a sealed network rather than being constantly discharged and replaced.
“ The water used in modern systems is essentially a top-up to a closed loop. It’ s not a continuous flow that is consumed and lost,” he noted. Because the water remains within the system, its quality becomes more important than its origin. Water treatment is required to prevent corrosion, scaling or biological growth within pipes and heat exchangers, but this does not necessarily require potable water.
One issue Piorko highlighted is that many data centres still rely on potable water for cooling even though it is not technically required. Treated non-potable water – such as reclaimed or industrial-grade water – can often be used instead.
In regions where water scarcity is a major concern, including parts of Africa, this distinction becomes particularly important. By shifting away from drinking-quality water, operators can significantly reduce pressure on municipal supplies.
“ The water needs to be treated because of the closed-loop system. But it doesn’ t have to be potable water,” Piorko explained.
AFRICA’ S EMERGING DATA CENTRE HUBS The discussion also touched on the geography of Africa’ s growing data centre market. Historically, the continent’ s major hubs have been South Africa, Nigeria and Kenya, driven by relatively mature digital economies and favourable regulatory environments.
Today, Piorko says the landscape is expanding. In addition to the traditional‘ big three’, Egypt and Morocco are emerging as significant new locations for hyperscale and colocation facilities. These markets are attracting investment thanks to improving infrastructure, strategic geographic positioning and growing demand for digital services.
At the same time, other countries may become attractive due to their energy resources – particularly hydropower. Piorko pointed to opportunities in countries such as the Democratic Republic of Congo, Rwanda and Uganda, where abundant hydroelectric capacity could support energy-intensive computing infrastructure. However, water and electricity availability are only part of the equation.“ When choosing where to build a data centre, it’ s not only about water and power. Regulatory stability, political conditions and the broader business environment are also critical,” Piorko said.
BEYOND WATER: DATA SOVEREIGNTY CONCERNS While environmental considerations often dominate the public discourse about data centres, Piorko argued that Africa faces another challenge that may be even more significant: data sovereignty.
Much of the data generated on the continent is currently stored and processed outside Africa, often under foreign jurisdiction. In addition, many of the major cloud platforms used by African businesses are owned by multi-national technology companies headquartered elsewhere.
This raises strategic questions about control, governance and the economic value of data. He suggested that Africa should develop its own strategic roadmap for artificial intelligence and digital infrastructure, including locally owned data centres capable of hosting and processing African data.
Ultimately, Piorko believes the public debate about data centres needs to be more nuanced. Water usage is a legitimate consideration, particularly in water-scarce regions, but modern cooling technologies have significantly reduced consumption compared with earlier designs. In many cases, the real challenge lies not in the amount of water required, but in how that water is sourced, treated and managed.
As AI drives the next wave of digital infrastructure investment, the focus may increasingly shift toward optimising resource efficiency while strengthening local digital ecosystems.
Data artificial centres intelligence
RACA Journal I July 2026 www. refrigerationandaircon. co. za 19