When our insatiable appetite for data can harm the environment too

Luis Rezende

SINGAPORE (CNA) – Guess what’s the most-viewed YouTube video of all time?

It’s the music video for hit song Despacito, which as of August 2020, has amassed a whopping 6.9 billion views.

While an admirable accomplishment, this came at a significant environmental cost. To power every search, click, or streamed video, six to eight data centres in different parts of the world must work together.

Since its debut in January 2017, Despacito has consumed more energy than 40,000 United States (US) homes do in one year.

Today, we still use terms such as gigabytes and terabytes but these will soon become antiquated because the entire digital universe is expected to reach 44 zettabytes by this year.

If this number is correct, it will mean there are 40 times more bytes than there are stars in the observable universe, according to a report by the World Economic Forum.

Even though most of us rely on the Internet daily, so few of us think about how this data is stored, and the energy needed to supply and capture such data.


Data centres form the backbone of the Internet – they store, process and communicate the data needed for any online service to work.

These warehouses comprise servers and cooling systems, which require electricity to run.

Collectively, data centres around the world account for around two per cent of greenhouse gas emissions – a level matched by the aviation industry.

By 2025, this figure is expected to reach 3.2 per cent, then to surge to 14 per cent by 2040 – the same as the US’ share of global emissions today.

Add to that the impact of COVID-19 which has driven people to shift to remote working, learning and play.

Networking tools like Microsoft Teams hit 75 million daily active users in April, a jump of 31 million from just over a month before.

Video-sharing social networking app, TikTok, was downloaded 351 million times within the first three months of the year – with total downloads up more than 100 per cent from the end of 2018. Within the same period of time, nearly 16 million people created a Netflix account.

The world has been forced to embrace digital transformation at unprecedented speeds and will continue to do so, as long as a vaccine is under development and social distancing restrictions remain.

We are seeing plans for “green data centres”, which are dedicated to minimising their carbon footprint. Tech giants such as Apple and Google are already investing in better environmental labels and standards, such as Fossil Free Data.

While these efforts are to be encouraged, they do not address the issue of dated infrastructure at existing data centres, which are computationally thirsty and hardware-heavy – making significant contributions to the carbon footprint of a company.

In information technology, the same output can be generated in many different ways – much like the same car model can be manufactured following different production methods.

Until recently, picking the least wasteful and energy-consuming methods was not a priority for many IT system designers. If those systems produced the results required from them, little attention was paid to whether they were doing so through the shortest and most efficient paths.

Excessive consumption of hardware resources – and of energy – often ensues.


Today’s software engineers have embraced a dangerous unbounded mentality that mounting more CPUs holds the answer to completing an intensive task. It is typically easier – and quicker – to go ahead with brute-force approaches, throwing raw computational muscle at a problem instead of redefining it around lower processing needs.

Writing more efficient software that can do the same with less hardware usage typically requires more work, in the form of deeper planning and technical articulation. Some engineers thus feel tempted to settle for more hardware-demanding code alternatives.

Seeing as IT infrastructure and services permeate every corporate function and department, chief information officers are in a unique position to champion and drive business sustainability within the organisation.

Unlike updating hardware and investing in new infrastructure, fine-tuning software engineering techniques comes at a much lower financial barrier. This starts with the strategic building and designing of software, leveraging engineering principles that deliver efficiency and high performance.

For example, choosing efficient data structures (to organise and manage data inside a program), avoiding unnecessary programming loops (which repeat the same instructions over and over again when the same outcome can be achieved with fewer repetitions) and deploying multithreading (in which CPU resources are more efficiently shared by a program) are basic techniques that can drive less overall hardware consumption while keeping end results unchanged.

While sustainability initiatives require an upfront investment, payback often comes in the form of cost-savings. At Niometrics, we’ve seen a 90 per cent reduction in hardware capacity and respective energy consumption among our clients after they replaced their older software platforms with our new ones.

A recent deployment for a large communications service provider will result in energy savings of 69,000,000 kWh over the course of five years – equivalent to the electricity 200,000 households consume in a single month in Singapore.

While their older software solution relied on 286 servers, our deployment delivered the same output with just 30 servers.


The digital revolution is at a crossroads. While there’s no denying the huge potential to keep transforming and improving lives, this can only come with consuming continuously larger amounts of human and machine-related data.