Tired after a long day and rushing into the couch with a comfy blanket and a show streamed on Netflix isn’t as short as it sounds. In the blink of an eye, the stacks of circuit boards probably one amongst the US, Mexico, UK, Canada, or Brazil are running to serve your comfy evening. Assuming that you are in a carbon-intensive country, by the time you have finished streaming the first episode of Squid Game, you have successfully contributed 100 gCO2e. As per the recently published data by Netflix, this is the carbon emission from streaming for an hour. Given its extraordinary usage during the quarantine, data shows that more than 6 billion hours of streaming was done every month by American subscribers only. The very basic calculation here amounts to the carbon footprint of 6 x 1011 gCO2e per month for streaming your favorite shows. For a better idea, a diesel car that produces about 121.5 gCO2e/km, will be required to travel 4.95 x 109 km to produce as much carbon emission as in a month of streaming in the US. That is equivalent to emissions produced during round trips from the Earth to the moon 64,386 times in your diesel car. Apologies to the Starship, bear with the comparison ;).
Carbon emissions from streaming for a month in the US during quarantine were equivalent to emissions produced during round trips from the Earth to the moon 64,386 times in your diesel car.
While the numbers presented above are not simply from the energy usage in data centers but a cumulative average of the emissions resulting from the entire process which involves the electricity usage of the device, device resolution, the system setup behind the internet connectivity to your place, and the data centers. On a global scale, data centers consume about 1% of the electricity demand and emit 0.3% of the annual carbon emissions. Previous data from several studies present even higher values of emission but the entire process of calculating the net emissions from data centers involves numerous assumptions taken from different sources which makes it a potential topic of detailed investigation on a national if not global level. The fact-checking of some of these studies as performed by IEA denies the validity given the current state of digital technology. A sigh of relief comes from the rapidly advancing technology to improve the efficiency of these servers combined with renewable energy-powered centers. The entire performance of data centers is driven by the equipped processors and as per a report by Nature, the computation power per kWh of a computer has increased by 10 billion times in the last 50 years. A round of applause for all the genius minds involved there!
One of the significant steps towards performance improvement of data centers is migrating to a hyperscale data center. As per Bill Carter, CTO at the Open Compute Project, a server in a hyperscale data center is capable of replacing 3.75 servers in any conventional data center. Although not an apples-to-apples comparison but a hyperscale facility by Google has its PUE 1.7 times lesser than the traditional data centers. The cumulative average of all its centers is 1.12 and the ideal PUE score is 1.0. Schneider Electric provides several TradeOff Tools to analyze the carbon footprint, power sizing, efficiency, and energy allocation of data centers. A few of the several factors that affect the net carbon emissions from these data centers involve location, efficiency, and load.
Like any other field, defining a system boundary is essential in calculating the carbon footprint of data centers. GHG emissions from all the facilities that run together to keep the data centers functioning are attributed to data centers. Four main driving factors in defining the system boundary are; direct electricity usage, water consumption during the production of electricity, water consumption in the cooling system, and wastewater treatment. As per an article in the Environmental Research, the 2018 annual operational water footprint (direct and indirect) of data centers in the US was estimated at 5.13 × 108 m3, and more than half of this is Water Scarcity Footprint. Water Scarcity Footprint as defined by ISO is the negative impact on social and environmental water supply when the available water for such demand is either polluted or used for producing goods and services.
As the world advances towards green energy, there is much hope that the increasing demand for computation power and electricity consumption will not depend solely on fossil-based power generation. Industrial giants like Facebook, Microsoft, Google, Apple, Amazon, etc with some of the largest data centers have already started their fight against climate change by migrating most of their operations to renewable energy but there still is a long way to go until we reach a carbon-neutral goal.
Regardless of what power and capacity these billions of circuit boards hold, they can not be traded for the only habitat planet we have. Conscious action is a way to go!