Watts next – can IT save the planet (and your bottom line)?

Vishal Barapatre, Group Chief Technology Officer at In2IT Technologies. (Image: Supplied)

---

The digital age’s insatiable demand for computing power has collided with an urgent and pressing need for sustainability. As data centres and AI workloads consume unprecedented energy, IT providers are pivotal in redefining how technology intersects with environmental stewardship. From optimising algorithms to re-imagining infrastructure, the path to energy-efficient computing is both a technical challenge and a strategic imperative – especially in markets like South Africa, where energy constraints and sustainability goals converge.

Traditional computing infrastructure presents a significant environmental challenge. Data centres, the backbone of the digital economy, consume an estimated 1%-1.3% of global electricity, a figure projected to double by 2026 as AI adoption accelerates and more businesses move their data and operations to the cloud. This substantial energy consumption contributes significantly to carbon emissions, often relying heavily on fossil fuels to power these facilities. The carbon footprint of data centres is now comparable to that of the aviation industry, raising concerns about their impact on climate change. Furthermore, these facilities often rely on extensive cooling systems to prevent overheating, which consumes vast amounts of water and contributes to potential water pollution. Ageing hardware also contributes to the problem of generating electronic waste (e-waste) – a growing concern globally, particularly in regions like South Africa, where improper disposal can lead to environmental contamination and ecosystem damage. Legacy systems, therefore, present a multifaceted environmental challenge that demands innovative solutions.

The exponential growth of artificial intelligence (AI) further intensifies the strain on energy resources. Training a single, large machine learning model can emit as much CO₂ as five cars produce over their lifetimes. This highlights the substantial energy requirements of developing and deploying complex AI algorithms. However, the good news is that AI itself offers a potential path towards more sustainable computing. Energy-efficient techniques like model pruning (trimming redundant neural networks to reduce computational complexity) and quantisation (reducing the precision of data used in calculations) can slash computational demands by up to 50% without significantly sacrificing accuracy. IT providers are increasingly deploying these and other innovative strategies to help their clients balance the ever-growing demand for performance with the critical need for sustainability. They are helping businesses optimise their AI workloads to minimise energy consumption while achieving their desired outcomes, offering a beacon of hope in the quest for sustainable computing.

To drive meaningful change, IT providers are focusing on three key areas:

1. Architecting green infrastructure

Transitioning to renewable energy sources is no longer a choice but a necessity. Forward-thinking IT providers are partnering with data centres powered by solar, wind or other renewable energy sources, drastically reducing their Scope 2 emissions. Scope 2 emissions are indirect emissions from purchased electricity, and by reducing these, companies can significantly lower their overall carbon footprint. Virtualisation, a technique that allows multiple virtual machines to run on a single physical server, also reduces hardware needs. By consolidating workloads onto fewer physical machines, organisations can lower energy consumption, reduce cooling costs and minimise their overall environmental footprint by as much as 70%. In countries like South Africa, where load-shedding and grid instability are persistent challenges, hybrid solutions that combine renewable energy sources with battery storage are gaining significant traction, providing a reliable and sustainable power supply for critical computing infrastructure.

2. Optimising AI for efficiency

Beyond implementing algorithmic tweaks, IT providers are actively redesigning AI workflows to maximise energy efficiency. For example, a South African financial institution successfully reduced the energy consumption of its fraud-detection model by an impressive 40% by employing knowledge distillation. Knowledge distillation is a technique that involves training a lightweight "student" model to replicate the behaviour of a larger, more complex model. This approach aligns with broader global trends, where sparse AI models that prioritise critical data points demonstrate the potential to reduce training energy consumption by as much as 30%. By optimising AI algorithms and workflows, IT providers can significantly reduce the environmental impact of AI-driven applications.

3. Circular IT practices

Electronic waste (e-waste) is a growing global problem, with a large share estimated to end up in landfills, contributing significantly to toxic waste. To combat this challenge, IT providers are implementing circular IT practices, such as hardware refurbishment programmes and modular server designs, which extend the lifespan of equipment and reduce the need for frequent replacements. In Cape Town, for example, a tech start-up is now repurposing decommissioned graphics processing units (GPUs) from crypto-currency mining operations for AI training. This innovative approach not only diverts e-waste from landfills, but also provides a cost-effective solution for AI developers. By embracing circular IT practices, IT providers can minimise waste, conserve resources and contribute to a more sustainable digital ecosystem, underscoring the necessity of these practices.

Regulatory pressures are increasing globally, pushing organisations to adopt more sustainable practices. For instance, South Africa's draft Climate Change Bill mandates emission reporting for large enterprises, creating a strong incentive for businesses to seek out IT partners that prioritise environmental, social and governance (ESG) compliance. At the same time, consumers are increasingly favouring brands with verifiable green credentials, creating market demand for sustainable products and services. IT providers offering carbon tracking integrated into their cloud platforms, enabling clients to monitor their emissions in real-time, are gaining a significant competitive advantage. By helping their clients meet regulatory requirements and appeal to environmentally conscious consumers, IT providers are positioning themselves as essential partners in the transition to a more sustainable digital economy.

The future of computing depends on finding a sustainable balance between innovation and environmental responsibility. For IT providers, this means embedding sustainability into every aspect of their service delivery model, from advocating for open source energy-efficient frameworks to designing AI as a service platforms powered by renewable energy. In countries like South Africa, where energy poverty and digital transformation co-exist, these strategies are not simply ethical considerations but also pressing economic imperatives. By proactively marrying cutting-edge computing technologies with circular economy principles, IT providers can transform sustainability from a perceived cost centre into a powerful catalyst for growth and long-term value creation. The industry has a vital role in ensuring a sustainable and equitable digital future for all.