Cape Town, South Africa –photo by @timalanjohnson

Water Scarcity, Data Centres and AI in South Africa

How Artificial Intelligence and Increases in Data Influences Water

Is water scarcity a threat to the digital economy or is the digital economy a threat to water scarcity? Let me explain: most servers need a great deal of water for cooling. There is a demand as well for a great deal of energy to run data centres. Humans of course cannot live without water, and currently large regions of the planet are increasingly drying up. That brings us to the question in the first sentence, because threats to water or to the digital economy is increasingly happening due to the massive increase in data being stored.

If we continue on this trajectory it is not only going to require a whole deal of energy. It will require alongside it a great deal of water, or other forms of cooling as well as the material cost of building servers.

In addition far more water is used to generate electricity that powers data centers than to cool them. It takes about 7.6 liters of water on average to generate 1kWh of energy in the US, while an average data center uses 1.8 liters of water for every kWh it consumes, according to the researchers.

In machine learning there is a want for more image and video data, and there is an increase in this unstructured data — not in a database, but more complex and in need of a greater deal of data storage. Artificial intelligence is often argued to require a great deal of data.

And where does that data go?

Let us take one example.

South Africa has been known to be hit by one of the worst droughts in a while, back in 2018 there was a fear that the water reserves would deplete, and if that would have been the case a large crisis would have resulted.

The things is… The world is not exactly getting cooler.

In fact even if we look at disagreement there is a somewhat clear consent.

An already fragile area will undoubtedly be, and is getting hotter.

Amongst this increase and scarcity —

cloud providers are entering South Africa.

“Microsoft has beat its biggest cloud rival Amazon by launching data centers in South Africa first. Amazon Web Services expects to bring its first data centers in the country online next year.”

There is a talk of hyperscale and job creation.

However data centres have not proven to bring much of either.

In computing, hyperscale is the ability of an architecture to scale appropriately as increased demand is added to the system.

If so, and brought to the outside world it is quite the contrary.

This seems less like hyperscale, and rather a lot more like hyperstupidity.

Why build a lot of data centres that need a great deal of water reserves in one of the most water scarce places — that have already been on the brink of a great water disaster?

It seem nothing but irresponsible to do so.

Yet I should not judge fully, as I would like to know the reasoning behind this.

Perhaps there is something that I fail to see beyond the clouds.

This is #500daysofAI and you are reading article 270. I am writing one new article about or related to artificial intelligence every day for 500 days. My current focus for 100 days 200–300 is national and international strategies for artificial intelligence.

AI Policy and Ethics at Student at University of Copenhagen MSc in Social Data Science. All views are my own.