If ChatGPT had written this 1 300-word article, it could have consumed more or less one cup of water. That’s according to a formula published in the Conversation by Leo Lo, an artificial intelligence (AI) literacy adviser at the University of Virginia in the US.

His calculations are based on ‘credible research’ showing that a 200-word GPT-5 response uses about 19.3 watt-hours and that a typical data centre uses 2 millilitres of water per watt-hour to cool its processors. Taken singly, it amounts to a drop in the ocean, but when you consider that ChatGPT responds to a total of 2.5 billion prompts a day, it becomes an ocean.

The irony is unavoidable – the point of this article being to illustrate the benefits of using AI or machine learning to manage Africa’s scarce water resources.

The Environmental and Energy Institute estimates the daily water consumption of a large data centre at up to 19 million litres, equivalent to the water use of a small town.

Offering a legal perspective, Carlyn Frittelli Davies, a natural resources and environment consultant at ENS, fleshes out the paradox in an online blog published in June. ‘As AI operations grow and the demand for water increases, especially for cooling large-scale computing facilities, […] regulations play a vital role in balancing technological advancement with environmental sustainability,’ she writes.

‘Conversely, the integration of AI may assist with water conservation when linked to services such as leak detection, irrigation optimisation, demand forecasting and public communication on outages.’

That said, Water Institute of Southern Africa (Wisa) CEO Lester Goldman lays out the uses for AI in the water sector in a recent Wisa blog.

‘AI isn’t just about robots and smart devices; it’s about using data and machine learning to make smarter decisions, faster,’ he writes.

‘In the water sector, this means identifying leaks before pipes burst, forecasting droughts more accurately, preventing water pollution and even improving how farmers irrigate their crops.’

For example, agriculture in South Africa accounts for close to 60% of all water use. He argues that using AI-powered irrigation tools (watering crops based on real-time data about soil moisture, weather and crop type) can cut water use by up to 60%, boost crop yields by at least 20% and reduce reliance on chemical inputs and manual labour.

His list goes on… ‘AI can also protect water quality. By analysing pollution patterns, machine learning models can detect contaminants from industries, mines or agriculture. In rural and under-resourced areas, where data is scarce, AI can still predict pollution risks, helping prevent health hazards before they reach communities,’ writes Goldman.

However, at the Africa Water Investment Programme (AU-AIP) Water Summit 2025 held in Cape Town in August, while acknowledging AI’s potential usefulness in managing and monitoring water supply, Yvonne Magawa, executive secretary of the Eastern and Southern Africa Water and Sanitation Regulators Association, cautioned that AI is only as good as the data it is fed… in other words, ‘garbage in, garbage out’.

‘If we don’t have reliable and accurate data, it doesn’t matter what we call it – AI or anything else – we will not get the results we want. Our job is to fix that foundation,’ she said.

The International Water Management Institute (IWMI) and Johannesburg-based Digital Earth Africa (DE Africa) are working to do exactly that, using satellite images and AI to deliver timely and accurate dam volume measurements, with the potential to transform reservoir management in Southern Africa.

Working with the Limpopo Water Course Commission, they have developed a digital twin of the Limpopo River Basin, a 400 000 km² area that takes in five countries – Botswana, Mozambique, South Africa, Tanzania and Zimbabwe.

The project is part of an initiative by the Consultative Group on International Agricultural Research to explore digital innovations that promote a water-secure future for Africa.

The virtual representation of the entire river basin now facilitates more accurate decision-making on water use, enabling water managers to visualise real-time data and to model data and watershed processes for forecasting on water availability and quality. The IWMI explains that uneven water monitoring capacity in the five countries in the basin had made it difficult to create an accurate hydrological model of the region.

‘New sources of data developed from a mixture of satellite images and machine learning go a long way to filling gaps in monitoring capacity,’ it says.

The comprehensive surface water datasets are derived from Landsat satellite images, and made available for free by DE Africa on a cloud platform.

‘This innovation shows how open access data can catalyse real-world impact, creating a way to track water availability in remote areas with minimal need for investment in data gathering, processing and field monitoring. With this data, the researchers could focus on developing methodologies that are now easily available for other users such as government water authorities, researchers and NGOs to adapt to more reservoirs and dams,’ according to the IWMI.

One of those trained in using the digital twin is Thulani Sibanda from Zimbabwe. ‘Using these AI tools makes life very simple for us. Sometimes we are working late at night trying to analyse the huge volume of data. Once you get used to these tools, it becomes more efficient to generate good reports for policymakers. These tools are necessary for us as a country,’ he said.

‘One of my responsibilities is monitoring irrigation areas,’ said Martha Zunguze of Mozambique. ‘We learnt that by using a digital twin, we can identify areas where irrigation is taking place in real time,’ she added.

The project hasn’t ruled out the use of citizen science to collect data.

‘We are very interested in biophysical data from citizens, but there is social information that is very valuable, such as indigenous knowledge,’ said Mariangel Garcia, IWMI’s research group leader for Water Futures Data and Analytics.

‘We need to support communities to better understand what technology – and specifically AI – can do for them. Then, we can be creative in together trying to provide solutions for our communities.’

The IWMI is, meanwhile, also involved in a project in the Middle East and North Africa to develop an AI-powered platform called e-ReWater, which uses remote-sensing technologies to monitor land, seas and the atmosphere. With help from satellites, it will estimate the availability of new water resources and improve water reuse.

AI is also being used in a more prosaic way in the water sector, by transforming the payment structure around water.

Speaking at the AU-AIP Water Summit, Alex Money, director of UK-based Watermarq, argued that AI could help the continent revalue water itself. He explained that existing tariff structures do not distinguish between rural communities who use small volumes for basic needs and industrial users that use enormous quantities but continue to extract high economic returns.

‘The reality is you need value-reflective tariffs to make the system sustainable. AI allows us to design shadow pricing models that capture the true economic and social value of water. This can mobilise investment, incentivise efficiency and support the financial sustainability of utilities.’

The company employs AI-powered models to collect and process large datasets – from satellite imagery and aquifer readings to climate projections – to set tariffs based on local realities.

Also at the summit, Ramateu Monyokolo, chair of Rand Water and the Association of Water and Sanitation Institutions in South Africa, said AI was poised to reshape how water is financed across the continent.

‘AI allows us to model risk in new ways, to create financial products that were previously unimaginable. We are seeing the beginnings of AI-driven credit assessments for utilities, which could make a significant difference in investor confidence,’ he said.

It’s undeniable that, as Wisa’s Goldman says, AI could be ‘a gamechanger for Africa’ … but there’s a caveat, only ‘if [it’s] deployed ethically and inclusively’.

Not only must it be beneficial for all citizens of Africa, it must also be deployed with thought to the sustainability of the digital infrastructure in a water-starved region.

Images: Gallo/Getty Images