Your email, online shopping, and back-to-back video meetings all generate data. The required data processing doesn’t happen just on your computer, it happens largely in the Cloud, and the Cloud is powered by data centres.
The pandemic has profoundly changed the way we live and work. Tim Berners-Lee, the creator of the web, states that Covid-19 has highlighted the fundamental importance of internet connectivity. So it’s perhaps no surprise that data centres have come to be regarded as the fifth utility, as critical as water, electricity, gas, and telecoms. Data centre staff have also been categorised as key workers.
Utility status may, however, be a double-edged sword. For one thing, governments will have higher expectations. The expansion of data centres also will need increased artificial intelligence (AI) processing capabilities, but if poorly designed and executed, these can have astonishing carbon footprints.
Revenues will approach $1tn as data volumes grow
The data stored in data centres globally will increase nearly five-fold in 2021 to reach 1.3 zettabytes (a zettabyte has 21 zeros), and the amount of data stored on devices is expected to reach nearly 4.5 times that at 5.9ZB. Much of this expansion will be driven by big data and internet of things (IoT) applications, both at home and in the workplace.
GlobalData predicts that Cloud service adoption will drive new data centre demands. In 2016 global data centre traffic reached an estimated 6 ZB. In 2018 Cisco’s Global Cloud Index forecast that global annual data centre traffic would reach 19.5 ZB by 2021. These estimates are from before we knew that millions of people would have to work from home during a pandemic, relying on video streaming collaboration tools to communicate with colleagues.
GlobalData estimates that the data centre market was worth $466bn in 2020 and that it will grow at a compound annual growth rate (CAGR) of 6.7% over the near ten years, reaching $948 bn by 2030. Much of the market expansion will come from the building of more hyper-scale data centres and the introduction of edge data centres. Estimates suggest that at the end of 2020 there were approximately 600 hyper-scale data centres.
The environmental cost of data processing
But data centre processing, especially for AI workloads, comes at a huge environmental cost.
Researchers at the University of Massachusetts performed a lifecycle assessment for training several common large AI models. The results were astonishing: training a single AI model can emit as much as 626,000lb of carbon dioxide equivalent – nearly five times the lifetime emissions of the average American car.
The data centre industry is under pressure to improve its sustainability record due to continued concerns about climate change. At the 26th UN Climate Change Conference of the Parties (COP26) in November 2021, eyes will be on data centres’ contribution to global warming.
The rapid growth of Cloud computing has seen the largest players deepen their commitments to green energy. Some data centre providers recognise that using renewable energy is now a prerequisite. For example, Google has committed to running its business entirely on carbon-free energy by 2030. Those that are unable to demonstrate their renewable energy adoption credentials will be at a disadvantage.
Data is the new oil, and it comes at a price
Both data and oil, once mined and refined, are highly lucrative commodities.
But unlike oil, the full cost to the planet of data mining and processing remains to be seen. Data centres will likely face future stringent sustainability targets – an unwelcome reward for an industry that excelled during the pandemic