Data Centres and Day Zero: UCT Panel Raises Alarm Over Water Crisis While AI Goes Unchallenged

Image: Taylor Vick

This July, the University of Cape Town held their fifth instalment of the Collective Conversations on Science for Society series, during which a panel explored the urgent need to rethink water. The discussion featured representatives from the university, the private sector, and the City of Cape Town. The panel centralised the theme of rethinking water use and waste in future cities, as well as making Cape Town a city conducive to water-sensitive design.

During the conversation, the panellists raised concerns regarding South Africa’s continuous grapple with deteriorating water infrastructure along with increasing urban demands. Among these concerns were those of depleting dam levels and the risks posed to informal settlements due to improper stormwater management systems. The panellists also took time to highlight the national Water Action Plan – a government strategy which focuses on addressing the failing local water infrastructure through investment, regulatory reform and the tackling of corruption within the sector.

One issue, however, remained relegated to the fringes of the discussion – data centres. Anxious to offer uninformed insights, the panellists opted to steer clear of the subject matter. The issue was eventually addressed, briefly, when one panellist, Dr Craig Tanyanyiwa from Water Footprint Implementation, suggested that South African citizens need to rethink their engagement with generative A.I. (Artificial Intelligence) technologies.

Dr Tanyanyiwa, on the City of Cape Town’s endorsement and integration of data centres in Cape Town, stated that individuals who still made use of A.I. assistants such as Claude and ChatGPT were “as culpable as the City of Cape Town.” Going on the offensive, he elaborated that the data centres exist solely due to a demand for A.I. assistance from South African citizens. The hypocrisy of the accusation made by Dr Tanyanyiwa stood in stark contrast to an earlier remark made by Deputy Vice-Chancellor for Research and Internationalisation Professor Thokozani Majozi. Professor Majozi offered that “we should create a huge space for asking: How does what we do in our little spaces affect the broader communities that surround us?” and “how do we change the life of the man down the street?”

The debate was precipitated by the proposed installation of two “massive, water-hungry and energy-hungry data centres next to Cape Town International Airport.” Journalist Steve Kretzmann reported that per year, a data centre using traditional cooling techniques consumes approximately 25.5 million litres of water. Kretzmann further details that these data centres, proposed by Equinix, would have an electrical demand of ~174MW — roughly 4.4 billion litres of water per year. The energy output required for the standard operation of the proposed data centre would be sufficient to power an area the size of Robben Island.

The proposal, by Equinix, notably lacks sufficient details regarding how much water the data centre will require, the estimated amount of energy needed, potential backup diesel generation, fuel storage, possible air contamination, noise pollution, and plausible health risks posed to the public.

Here is where Professor Majozi’s remark comes into play; UCT’s newly minted A.I. policies do not look out for the proverbial ‘little guy.’ Last year, the university introduced an A.I. framework which will oversee the operationalisation of A.I. in teaching and learning spaces (such as coursework and assessment) as well as utilising the technology across systematic processes.

A study conducted by Miguel Yañez-Barnuevo of the Environmental and Energy Study Institute revealed that each 100-word prompt uses approximately one bottle of water/519 millilitres. Therefore, an institutionally backed initiative encouraging —even forcing students to utilise generative A.I. across coursework would undoubtedly place undue pressure on already strained water reservoirs.

UCT says that these policies are actually an attempt at ‘catching up to’ other universities who have already implemented A.I. into educational frameworks. It is then far from conjecture that a nation-wide restructuring of tertiary curricula, around the necessitated use of generative A.I., threatens water availability for ordinary South African citizens who rue the possibility of another Day Zero.

The Collective Conversations panel additionally emphasised that South Africa is notoriously a water-scarce country, with water insecurity likely to worsen. They highlighted that 80% of municipal water infrastructure is in poor or critical condition, escalating the water crisis to a “severe infrastructural and socioeconomic threat.”

The Dam Levels Report issued by the City of Cape Town on the 3rd of August 2026 detailed a 20.7% decrease in water levels across major dams, with total capacity in 2024 at 99.3% falling to 78.6% in 2026. The report also detailed that exceptional rainfall during May 2026 resulted in a remarkable recovery for the Western Cape Water Supply System (WCWSS), increasing the likelihood of the system reaching full supply capacity by November 2026.

The WCWSS dam storage levels, however, also advise the public to practice wise water use, and that dam levels are nearing “early drought caution.”