The urgency of addressing the climate crisis underscores the necessity for collective action, with businesses playing a pivotal role in reducing emissions to achieve net-zero targets. In recent years, there has been a notable increase in awareness regarding the environmental impact of technology firms. A significant aspect of this impact lies in the energy consumption of data centres and data transmission networks, which collectively contribute between 1% to 1.5% of global electricity consumption, as reported by the International Energy Agency.
However, attributing emissions solely to data centres would be incomplete. Another significant source lies in equipment manufacturing. For instance, Greenpeace East Asia forecasts that by 2030, the electricity consumption for semiconductor manufacturing alone could reach 237 terawatt hours. This amount is staggering, equivalent to the entire electricity consumption of Australia in 2021.
Manufacturing emissions extend beyond the infrastructure within data centres, including servers, networking equipment, buildings, and cooling facilities. End-user electronic devices, such as computers, mobile phones, laptops, and networking equipment used by employees, also play a significant role in carbon emissions. As these devices often come with a hefty price tag, most companies already have effective lifecycle management policies for this equipment, which would also result in reduced carbon emissions.
However, there's a frequently overlooked contributor to carbon emissions in IT systems: Software. Although this might come as a surprise initially, considering that hardware requirements are heavily influenced by software platforms, it makes sense that improving software efficiency could result in decreased hardware usage. This, in turn, would likely lead to a reduction in emissions. In essence, optimising software performance plays a significant role in minimising the environmental impact of IT operations.
The most widely used approach for measuring, managing, and reporting emissions from operations and value chains is the framework developed by the Greenhouse Gas Protocol. The protocol includes different levels of guidance:
● Scope 1: Direct emissions stemming from an organisation's activities.
● Scope 2: Emissions attributed to energy consumption within the organisation.
● Scope 3: Indirect emissions originating from sources such as the supply chain, among others.
Emissions arising from an organisation's direct utilisation of cloud services fall under the category of Scope 3 emissions. Historically, quantifying emissions from cloud usage has been a challenge, with many of these emissions going unmeasured. However, with influential companies such as Amazon, Microsoft, and Google now transparently disclosing the emissions associated with their cloud services, the pressure is mounting for users to follow suit and accurately measure their own Scope 3 emissions. This shift towards greater transparency and accountability is driving a broader recognition of the environmental impact of cloud computing across various industries.
Developing a Sustainability Strategy
A comprehensive sustainability strategy should advocate for:
● Embracing the consumption of sustainable energy sources.
● Enhancing the efficiency of data centres.
● Utilising efficient end-user electronic equipment.
● Implementing and deploying more efficient software platforms.
As discussed, there has been increased attention on the first three elements in recent years. However, leveraging a more efficient software platform can be equally impactful. For instance, let's consider the role of databases:
In a comparative analysis between our database platform and a competitor's product, we discovered that transitioning to a more efficient platform could lead to a remarkable 80% reduction in carbon emissions. This finding underscores the significant environmental benefits of efficient software. Moreover, adopting such software solutions not only reduces resource consumption but also proves to be more cost-effective, enhances performance, and simplifies maintenance processes. This multifaceted advantage highlights the compelling case for prioritising the adoption of efficient software in sustainability efforts within the technology industry.
This strategy is gaining traction among numerous companies. Take TomTom, a prominent location technology developer recognised for its in-car mapping and navigation systems. TomTom's migration to Aerospike resulted in an impressive 86% decrease in carbon emissions from its platform while simultaneously enhancing performance and stability.
The Evolving Importance of ‘Efficiency’
With the emergence of Large Language Models and the need for ever-increasing computational and storage capacities, businesses face a significant sustainability challenge. For instance, researchers at the University of Massachusetts, Amherst revealed that training a single AI/ML model can emit over 626,000 pounds of carbon emissions – nearly five times the cumulative emissions of an average car over its lifetime. Hence, the efficiency of the underlying software infrastructure is poised to grow in significance.
Historically, software efficiency has often been equated with performance and speed. In the realm of computing, an algorithm can only outperform others if it utilises fewer resources (such as CPU cycles) to complete a task. Consequently, it follows that more efficient software necessitates fewer hardware resources to accomplish the same task. With mounting pressure on IT companies to comprehend the true impact of their technology, assessing the efficiency of software platforms has become indispensable.
Designing a Sustainable Future
We stand at the intersection of technological innovation and environmental responsibility. The path to reducing our carbon footprint is within our grasp, but first, we must fully comprehend the impact of technology on emissions, accurately measure and predict these emissions, and then take decisive actions to mitigate them.
One potential approach is for the software industry to adopt a standard that quantifies and publicises emissions associated with software platforms, similar to the energy efficiency labels found on electronic devices. This metric could serve as a non-functional requirement for assessing software platforms, empowering decision-makers to optimise their architectural choices with sustainability in mind.