The carbon cost of data

Clive Humby is credited with coining the phrase “Data is the new oil”. A mathematician by training and the marketing brain behind Tesco’s Clubcard, he observed in 2006 that “like oil, data is valuable, but if unrefined it cannot really be used.”

Now environmental scientists are beginning to worry that data resembles oil in its harmful effects on climate as well — not data per se, but servers, data farms and other hardware required for creating, storing and processing data for the increasingly digital life.

The world has been steadily getting interconnected and digitalised since the world-wide-web came into being. The rapid evolution of the Cloud, high speed Wi-Fi, and internet of things, and the development of digital durables, which are replacing their older analogue counterparts, have increased the quantity of data being generated, stored and analysed exponentially.

The trajectory of digitalisation moved up even more sharply when the pandemic struck. Companies were forced to increase their digital footprint. They also moved a lot of their operations to the cloud — which meant moving applications and data to huge server farms. According to Statista, the information/data created, captured, copied and consumed worldwide has grown from 41 Zettabytes in 2019 to 79 Zettabytes in 2021. By 2025, Statista estimates it will be 1881 Zettabytes.

Data stored in servers have a not inconsiderable carbon footprint, though. As with many things to do with the environment, however, it is hard to calculate precisely given the number of variables involved. Server farms consume a lot of electricity. Apart from the power required to run the servers themselves, the air conditioning systems needed to run them also guzzle a lot of energy. Data farms generate enormous heat and they need very powerful air conditioning systems to maintain optimum temperature. Then there is the water consumption. Studies have shown that data centres tend to use up water resources and heat the cities where they are located, despite all the air conditioning.

The carbon footprint depends on the power source it uses — and also the amount of air conditioning it requires. Many data firms still primarily depend on thermal or gas-fired power stations though many new ones are being built to utilise mostly renewable power.

The kind of silicon chips used has a role too in the power consumption. Older, more power-guzzling chips have a worse carbon footprint than those that are new and more energy efficient. Finally, other variables like how cold is the geographical area where the server farm is placed as well as the age and energy efficiency of the building all matter in the actual energy being consumed. Also, because Cloud servers need to offer zero downtime, they work with built in redundancies, which increase their energy requirements. Then, the buildings themselves are considered large emitters — especially buildings built in the past half a dozen decades. Older data farms are often in older generation buildings and that contributes to energy requirements too.

It is estimated that the cloud industry consumes more power than many nation states. And it is a bigger emitter than the global aviation industry.

The problem is likely to exacerbate. One, the pace of digitalisation has increased. And more data is being generated and captured than ever before. When Web 3.0 replaces the current Web 2.0, the consumption of data will increase further. This is simply because the blockchain technology on which Web 3.0 is being built consumes gargantuan quantities of energy.

Also, much of the data today is stored in cross-border servers, but as privacy concerns grow, the need for data to be stored locally is also likely to become a significant factor. A number of countries are already mandating that all personal data of their citizens be stored locally — or at least that a copy is stored locally rather than in some server farm in a remote country. This will lead to a huge spurt in data servers being built in every country, apart from the big global cross-border data farms.

India too is contemplating data localisation — though so far only financial transaction data needs to be stored locally. If India does mandate that all personal data and transaction data of its citizens need to be stored locally, it will require enormous server farms to be put up inside our borders. India is the second largest data generator in the world, just behind China.

Data centre companies have tried many things to reduce their consumption of energy. They have built data centres in cold regions but, in general, it has not proved very practical. They have also tried building submerged data centres in the sea to ensure they remain cool without the need for air conditioners. But then again, the ecology of the sea changes when the data centres are placed under water. It adversely affects marine life. Now engineers are trying to build more energy efficient chips, use green software and generally make data centre operations more energy efficient while also using renewable sources, rather than thermal-fuelled electricity. Big Tech — Google, Amazon, Microsoft, Meta etc — have also pledged stiff energy and carbon reduction targets.

The conversation around the environmental footprint of information technology, data, Cloud, the internet and other digital technologies is slowly growing. India needs to join the discussion too.

The writer is former editor of Business Today, Businessworld, and founder of Prosaicview, an editorial consultancy