By Behrad Babaee
It will be of no surprise to anyone that the energy consumption of computers is growing quickly. According to research, in 2020 the Information and Communication Technology (ICT) sector, which includes data centres, networks and user devices, hoovered up around 915 TWh of electricity – 4-6% of all electricity used in the world – but this is set to rise and could even reach 3,200 TWh by the end of the decade. Even more startling is the news that the cloud computing sector now has a carbon footprint larger than that of the airline industry.
What puts further demand on energy resources is the widespread adoption of Artificial Intelligence (AI) and Machine Learning (ML). Ingesting, deploying and applying data globally to serve the real-time needs of customers means pushing data out to the edge and increased pressure on cloud servers.
Balancing the good and bad of AI
While AI brings undeniable advantages in helping us battle against climate change, there are just as many disadvantages in terms of the toll its development is taking on our environment. Research conducted at the University of Massachusetts found that training several AI models can create a huge 626,000 pounds of carbon dioxide, while across its entire lifecycle a model will need feeding, tuning and support with hardware to serve its algorithms.
Anyone working in the technology industry should be aware of, and feel some responsibility for, ensuring we make our software and systems as sustainable as possible. Fortunately, there are ways we can do this without compromising on performance, whether that’s through coding efficient programs or maximising data centre efficiency. In fact, database technology provides a great opportunity to reduce server footprints which not only cuts emissions but can also reduce latency and enhance throughput.
For organisations it’s no longer possible to take the view that sustainability is a ‘nice-to-have’ but only if it doesn’t impact the business. The pressure on companies, not just to commit to Net Zero emissions targets, but to proactively meet them, now means that achieving a balance between reducing resources and energy consumption levels while increasing profits is essential. A study recently conducted by Climate Impact Partners, a leading global carbon finance organisation, found that companies that reduced carbon emissions earned approximately $1 billion in profit compared with those that didn’t.
Where sustainable software fits in
Until recently, operating software platforms were largely ignored when it came to accounting for carbon emissions. However, the considerable emissions that are being created by the expansion of IT infrastructure to support AI and ML applications means that IT decision-makers must factor in the environmental impact of their platforms. Software with efficiency built in will complete tasks more quickly and require fewer CPU cycles, and in turn will demand less hardware and energy, resulting in lower costs.
At the very heart of IT infrastructure is the database, which, if it has been designed with a focus on efficiency, can result in organisations requiring up to 80% fewer servers, while offering greater scalability, higher throughput and reduced latency. This can result in clients achieving 80% lower costs.
Make no mistake, carbon footprints are growing for multiple reasons, not least the adoption of AI and ML. The impact of this cannot be ignored. Instead, organisations must take responsibility for the financial and environmental costs of our technology usage and look for solutions that use efficient coding practices and are optimised for the data centre of the future. Groundbreaking data technology allows companies to revolutionise how they approach the IT stack, embracing levels of efficiency that will deliver a sustainable future.
Babaee is a Technology Evangelist at Aerospike