AI is now a fixture in the digital landscape, and its impact on global data centre energy consumption continues to intensify. Data centres, at the heart of AI infrastructure, are seeing a dramatic rise in power and resource demands. The latest International Energy Agency (IEA) research explains that data centres consumed between 240 and 340 terawatt hours (TWh) of electricity in 2022 alone, representing about 1-1.3% of global electricity demand.
The explosive growth of AI models, such as OpenAI’s GPT series, is putting unprecedented strain on data centre resources. This includes not only electricity but also water, which is used in cooling and energy production processes. And we all know overuse of water is a critical issue for society. As AI continues to proliferate across digital platforms, its energy consumption is growing by 20-40% annually, with tech giants like Amazon, Microsoft, Google, and Meta accounting for a large portion of this impact.
Reports highlight these concerns. A recent joint investigation by the Guardian and SourceMaterial found that the rising water demands of data centres are increasingly creating severe environmental challenges, particularly in areas facing water scarcity. Meanwhile a new IEA (International Energy Agency) forecast has predicted that energy demands from AI will quadruple by 2030, further exacerbating the pressure on the global power grid. Clearly, there is a need to address the intensifying resource demands of AI while mitigating its environmental impact.
The question remains: how can we efficiently resource this growing demand without harming the planet? In this article, I will explore how an emerging category of IT called ‘platform engineering’ can play a part in managing these demands while contributing to more sustainable data centre operations.
What is Platform Engineering?
Platform engineering is a modern approach to building and managing infrastructure that helps organizations meet operational goals efficiently. It does this by combining software development with infrastructure management, using automation, tools, and best practices to create systems that optimize resource allocation, improve efficiency, and enhance the productivity of developers.
A platform engineering approach focuses on creating a highly flexible, self-service infrastructure platform that allows development teams to access and manage resources easily. This means workloads can be scaled effectively while minimizing inefficiencies. By integrating platform engineering into operations, companies can ensure their data centre infrastructure is as efficient and sustainable as possible.
This is particularly crucial for AI, where the demands of training models are substantial. These models require enormous computational power to train and generate outputs, which in turn drives up energy and water usage in data centres, often in regions already facing water scarcity. The SourceMaterial report identified 38 active data centres owned by the big three tech firms in parts of the world already facing water challenges, with 24 more under development.
While platform engineering can certainly help reduce the environmental footprint of AI, it should not be seen as a catch-all solution. Achieving sustainability in the age of AI will require a combination of strategies, including critical policy changes, advancements in cooling technology, shifts towards renewable energy, and more efficient hardware. But it can help…
How platform engineering can help drive sustainability
Businesses can take an important step to mitigate the environmental impact of AI by improving efficiency across the entire lifecycle. Here are five ways platform engineering can help resource the growing demand for AI in data centres while addressing environmental concerns:
1. Intelligent Resource Management
AI workloads are energy-intensive, particularly during the training phase. Platform engineering tools help manage resources dynamically by providing real-time insights into energy usage and workload allocation. By scheduling AI tasks during off-peak energy hours and optimizing the use of renewable energy sources like solar and wind, platform engineers can reduce the strain on data centres and ensure more efficient resource utilization.
2. Scalable Infrastructure with Elastic Resource Allocation
AI systems, particularly large-scale models, require vast amounts of computational power. One of the biggest sources of wasted energy in data centres comes from idle resources. Platform engineering can enable elastic resource allocation, meaning computational power can be scaled up or down based on real-time demand. This reduces the energy wasted on idle servers and ensures that data centres are operating at peak efficiency.
3. More efficient cooling systems
Cooling is one of the most resource-intensive processes in data centres, often requiring significant water and energy usage. Platform engineering can play a role in integrating automated cooling systems, ensuring that cooling processes are optimized based on real-time data. By scheduling AI workloads during cooler parts of the day or using advanced cooling techniques, such as liquid cooling, companies can reduce both water and energy consumption.
4. Real-Time Monitoring for Resource Optimization
Real-time monitoring, often achieved through observability tools, is a core principle of platform engineering. By tracking energy consumption, water usage, and other resource metrics, platform engineering helps organizations identify inefficiencies and make informed decisions about resource management. By ensuring that AI workloads are routed to the most energy-efficient servers and data centres, platform engineering can optimize the overall sustainability of operations.
5. Integrating Environmental and Financial Goals
Platform engineering can integrate with frameworks like GreenOps (for sustainable practices) and FinOps (for financial operations) to align environmental goals with business outcomes. This allows companies to make data-driven decisions about energy usage, prioritize energy-efficient configurations, and ensure that their sustainability efforts are financially viable. This integration can lead to more sustainable practices across all levels of AI deployment.
A step towards IT sustainability
There is no silver bullet for the environmental challenges posed by AI’s rapid growth, and tech firms and data centre developers will need to factor in a mix of energy sources, technology adoption and sustainability strategies. The energy demands from AI and the rising water consumption in data centres will require a multifaceted approach.
However, as the AI landscape continues to evolve, platform engineering has a clear role to play in making sure data centres keep up with the growing demands of AI technologies, without compromising on sustainability. With the right strategies in place, platform engineering can help businesses navigate the challenges of powering AI in a way that supports both innovation and environmental responsibility.
