Data centres require a constant stream of power all day, every day of the year, which equates to 1% of the global energy demand1. There has been a steady increase in the reliance on data centres in recent years for a variety of reasons. Some of these reasons include the global shift to work from home2, increased data generation3, and AI (Artificial Intelligence) related workloads becoming more widespread4. All of this culminates in data centres seeing much more use, which in turn increases power consumption. The environmental impact of such power usage cannot be ignored as data centres contribute 0.3% of global carbon emissions. Regrettably, the other side of this foreboding coin is hardware and its disposal. 89% of organisations recycle less than 10% of their IT hardware, which results in c opious quantities of e-waste. 53.6 million tons of e-waste was generated in 2019 which is an increase of 21% in five years5.
The trend of increasing reliance on data centres is unlikely to change as the internet is becoming an indispensable part of everyday life. In June 2019, 58.8% of the global population (4.5 billion) used the internet – this increased to 62% (4.8 billion) in June 2020, and to 64.2% by December6. Keep in mind that this is just the number of users; internet traffic on the other hand increased by 40% in 2020 due to video conferencing, streaming, and online gaming according to TeleGeography's Global Internet Geography7. To maintain data centres, an estimated 200-terawatt hours (TWh) is needed, which is more than the national energy consumption of some countries in their entirety, such as Spain.
U.S. research firm, International Data Corporation (IDC), estimated that there will be a 146-fold increase in global data generation between 2010 and 2025. 1.2 Zettabytes (1.2 trillion Gigabytes) of information was created in 2010 – a 50% increase from the previous year. It was estimated back then that by 2020 the amount of data generated globally will be 35 Zettabytes. That was reached in 2018. 2020 created 59 Zettabytes, and it is predicted that by 2025, this number will rise to 175 Zettabytes8. This vast amount of data is often stored in the cloud, the reality of which means that most of it is stored in data centres, where it is kept safe on hard drives. Hard drives are made of mainly aluminium, glass, or ceramic, with a coating of CoPtCr (cobalt, platinum, and chromium), with traces of boron or tantalum. All of these must be mined, processed, and shipped among other supply chain activities. This adds more levels of complexity when tackling sustainability and eco-friendliness.
The posterchild of the 4th industrial revolution is Artificial Intelligence. AI is being used in a wide range of industries, like healthcare, all forms of research (e.g. physics, chemistry, biology, etc.), finance, and the l ist goes on9. The application of AI is becoming increasingly relevant as it can sift through unfathomable amounts of data faster and more accurately than any human could. This brings the benefit of unparalleled assistance in decision making and predictions. Working with AI requires two things: High-Performance Compute (HPC) systems, and data. There is most certainly no shortage of data, and computers and their components are becoming more powerful year on year. Such incredible compute requires proportional amounts of energy. In other words, the more powerful the component, the more energy it requires. Current HPC CPUs use around 800W. Even before the energy crisis, environmentally conscious suppliers have been finding ways to produce more efficient components that get more work done with less energy. For example, just a few years ago AI model training took days to complete but with current advancements training times have reduced to a matter of hours. This means these systems run at high intensity for shorter periods. With AI’s limitless potential, the reductions to environmental impact could be huge.
2019 produced 53.6 million tons of e-waste5. This includes everything from cables through motherboards, to chassis, and everything in between. The amount of e-waste is expected to grow to 74 million tons by 20305. According to Circular Computing 272 million new laptops are manufactured each year, 160 thousand “old” laptops are disposed of every single day in the EU – 70% of which are reusable. All this physical waste is having a terrible impact on our environment, but what makes this worse is the amount of raw material needed to manufacture a single laptop. To produce a single unit, the supply chain emits 316 kg of CO2 and uses up to 190,000 litres of water10. The amount of e-waste per capita varies by country, for example in 2019 the U.K. produced 23.9 kg11, the U.S. and France both produced 21 kg.
Going forward, waste disposal will have to be given an overhaul from the ground up as well as including plans on how to clean up previous years’ wastes. This is such a deep-rooted issue that change will need to be brought about by government.
Big Data and Business Analytics (BDA) and their solutions have seen a continuous growth in annual revenue which further proves that this sector is not going away any time soon. In 2019, the BDA saw a global revenue of $189.1 billion, which is a 12% increase on 2018. The Worldwide Semi-annual Big Data and Analytics Spending Guide from International Data Corporation (IDC) project that by the end of 2022 the global BDA revenue will reach over $270 billion12. With increasing reliance on data centres and IT, we need to realise that current infrastructure needs updating if more pressure from larger workloads is to become the norm. Currently, approximately 200 TWh a year is used to maintain data centres. This is only going to go up. The level of environmental impact caused by data centres is uncertain if upscaling carries on. Considering that there is room for improvement, on a larger scale, their impact will become more noticeable. Considering how much data centres are being used, making them more sustainable would carry a noticeable benefit.
To lower the carbon footprint of the IT industry, tech giants have set themselves agendas to lower their carbon emissions to net neutral, and even net negative. For example: Google who has been carbon neutral since 2007 and aims to run all their data centres on carbon-free energy by 203013. Microsoft aims to be carbon negative by 203014, which would mean they would be removing more carbon from the atmosphere than they would emit. Apple too aims to be carbon neutral across their entire business and manufacturing supply chain by 203015.
There are plenty of different and interesting approaches to become more echo friendly, such as supplementing data centre power demands with renewable energy, lowering energy consumption, turning waste energy into usable, or straightforward tree planting initiatives.
To become carbon neutral is to make changes in not only infrastructures, but also optimise technologies to make best use of what we have left.
As great as these steps are, government level changes will have to be made to see a much more substantial and quicker turnaround. For the past 150 years human knowledge has been rapidly developing, benefiting our welfare, at the cost of the environment. Now this knowledge must be used to solve problems without making more on the way.
Keep an eye on our blogs to find out more about the current approaches to sustainability.
To help our clients make informed decisions about new technologies, we have opened up our research & development facilities and actively encourage customers to try the latest platforms using their own tools and if necessary together with their existing hardware. Remote access is also available
A summary of what to expect from the RUN:AI event