South Africa’s need for digital infrastructure is growing. Digital transformation and digitisation trends are reliant on the rollout of infrastructure, including, most notably, data centres, writes Sienna Cana, global enterprise segment marketing manager at Axis Communications.
Substantial investments in new facilities by operators such as Microsoft illustrate the growth of the local market, while the global market is projected to be more than $256 billion (R4.6 trillion) in 2024.
Much of that infrastructure powers the day-to-day services that the world has come to rely on, but a huge part of the growth in demand comes from the next generation of service provision: the world of AI. Generative AI (GenAI) has received the attention it has thanks to its rapid evolution and adoption by businesses.
According to the South Africa Generative AI Roadmap 2024, published by World Wide Worx, Dell Technologies, and Intel, large enterprises in the country are embracing GenAI with many already using or planning to integrate it within their organisations.
However, predictive AI – designed not to generate data but to analyse and draw conclusions from it – has received a more muted public reaction. This is despite its potential to extract valuable insights from sound, images, and most importantly, video, beyond anything we could ever hope to achieve with humans alone.
AI is not just a growth driver; it’s a growth enabler. As data centres expand in size and become more complex, and as their locations spread to meet demand across the globe, AI will play a vital role in simplifying the local and remote management of data centre sites.
As power draws increase – GenAI alone is expected to require an additional 38GW by 2028, all while South Africa continues to face ongoing power supply constraints – AI will help find new efficiencies and discover sources of waste. As data centres enter their critical entity era, AI will play a crucial role in supporting the essential security and safety functions of these sites.
Video data is now a rich resource for AI analysis. A camera is potentially the strongest sensor a business could employ, generating millions of data points multiple times every second. Every pixel can be isolated and analysed, a single camera view split into numerous points of interest to allow one camera to perform multiple jobs at once.
Object-based analytics can detect, track, and classify items within a scene, and trigger automated processes based on easily defined rules. Cameras are versatile and their applications are almost limitless.
If a camera can see something, AI can act on it. Through deep learning, it is possible to develop custom reactive applications that offer new solutions to old problems or identify new problems before it is too late to act. And, contrary to the heavy AI workloads that underpin the rapid growth of data centres, properly trained AI models allow such analytics applications to run directly on the network edge, within the very camera hardware they rely on.
That means a camera already in use for security could enhance its abilities, using AI analytics to identify unauthorised personnel in sensitive areas and automatically sound the alarm, or detect and alert operators to suspicious activities like loitering or break-ins. But it also means that same camera could do more – it could integrate with an access control system to detect tailgating, or work in tandem with a thermal camera to offer operators a live view of any hot spots and even automatically trigger additional cooling.
AI’s creative potential means analytics applications can be moulded to fit the unique needs of the data centre environment. Object detection, for example, might be tuned to seek out banned items like water bottles. Cameras can be configured to detect visual or, through their microphones, audible signs of server failure or degradation. Analytics can be trained to watch for environmental hazards like leaks and ensure that upkeep and maintenance are adequate to prolong the life of equipment.
As data centre customer numbers grow, video analytics can offer co-locating clients visual verification of the precise status of their physical servers or help to optimise energy use through automatic lighting and cooling systems based on detected occupancy. Even disaster recovery can benefit from AI analytics. A camera detecting smoke could automatically trigger loudspeaker alerts, while cameras, intercoms, and readers catalogue the precise numbers and location of personnel to streamline evacuation procedures.
Data centres are the cornerstone of tomorrow’s technology, but nobody is saying that the rapid expansion of digital infrastructure will be easy. Operators in South Africa need every advantage they can get, be that saving money, saving energy, or just running facilities as cleanly, efficiently and safely as possible.
AI analytics offer all of these advantages and more, all as an extension of hardware, which would be required for the security functions regardless of whether analytics are used.