Can AI Save AI? The rise of green data centres

The AI revolution is reshaping the world, but beneath the surface of this transformation lies a pressing challenge of energy consumption & sustainability. The data centre industry, already responsible for around 1-1.5% of global electricity demand, is now facing an exponential surge due to AI workloads. As per IEA, data center energy consumption shall nearly double to 1000 TWh by 2026 from that of in the year 2022. This will have a corresponding impact on carbon footprint (greenhouse gas) and other resource requirement mainly water.

The question now is: how can the industry balance the need for AI-driven innovation with environmental responsibility and better manage energy, environmental, societal, and sustainability parameters associated with operating data centers? 

Being sustainable is not only an environmental & societal necessity but also crucial for business sustainability.

The answer lies in AI itself. Artificial intelligence is not only driving demand but can be the key enabler of sustainable and efficient data centres. Modern data centres can optimise energy use, improve cooling efficiency, reduce downtime, and minimise water consumption by leveraging AI-powered automation and monitoring tool with, machine learning, and predictive analytics features. AI is no longer just a workload housed within these facilities consuming energy & resources but also evolving into the intelligence that manages them in operational efficiency, assets management & advanced monitoring.

AI-Powered energy optimization & use of renewable energy sources 

Energy efficiency has long been a challenge for data centres, with a significant portion of power being lost in underutilized hardware in inefficient cooling & redundant infrastructures provided for increased availability. Traditional data centres operate at a low utilization ratio, often consuming excessive energy even at low work load as idle ITC infrastructures consumes 40% of power that of at peak work load. AI is transforming this paradigm by enabling real-time energy management, increasing utilization ratio & efficiency.

Google’s DeepMind AI, for instance, has demonstrated the potential of AI-driven energy optimization. By dynamically  autonomously adjusting cooling and power systems in real time, it reduced Google’s data centre cooling energy consumption by 40% leading to 15% overall energy saving in data centers. The system have cognitive capability to improve continuously by analyzing historical data.

Similar AI-focused models are now being deployed by Meta, Microsoft & Amazon across the industry, in the cloud. They are dynamically shifting workloads to different availability regions - saving energy & increasing sustainability by using renewable energy.

Transitioning to renewable energy sources is essential for the sustainable growth of data centres. Goldman Sachs report also suggests that renewables could supply up to 80% of the energy demand for data centres. Companies like Google and Microsoft have committed to powering their operations entirely with renewable energy, setting a benchmark for the industry.

Some of the countries of European commission are creating legal frame work for mandatory compliance of renewable energy factor. German energy efficiency guidelines requires data center operators   to cover 50% of their electricity consumption from 1 January 2024 and 100% from 1 January 2027 from renewable sources. 

While these investments are a step in the right direction, the availability of renewable energy can be quite inconsistent, posing challenges for data centers as well as electric grids. While innovations are being worked on to address these issues and minimize risks by integrating these sources with advanced energy storage solutions and grid management to address intermittency challenges, for developing countries like India, we also have the option of harnessing nuclear power.

Nuclear power is a reliable source of clean energy and has a minimal carbon footprint, making it a great green alternative. Use of AI can help advanced planning for proper mix of various energy resources based on forecasting energy demand, workload & weather conditions and select energy sources with lowest carbon footprint and helps data center to obtain carbon neutrality.

AI-Driven Cooling Systems

Cooling remains one of the most energy-intensive aspects of data centre operations, often accounting for up to 40% of total power consumption. AI is revolutionizing cooling efficiency by replacing static, one-size-fits-all approaches with dynamic, adaptive controls, predictive optimization and real time monitoring.

AI assisted liquid cooling, for example, can adjust fluid flow dynamically based on real-time heat loads, reducing power consumption of cooling system up to 80% compared to traditional air-cooling methods. Machine learning models are also being used to predict thermal hotspots before they occur, allowing for proactive cooling adjustments that prevent overheating while reducing unnecessary cooling cycles.

These advancements are particularly crucial as AI-driven hardware, such as GPUs and TPUs, generates significantly more heat than traditional CPUs. Without AI-powered cooling systems, the energy demands of AI workloads could spiral out of control.

Increasing workflow efficiency & data center operations through AI enabled software solutions

One approach to reducing energy consumption in data centers is optimizing workflow, task management, and efficiency. AI enabled Data Center Infrastructure Management (DCIM) Systems can provide real-time insights on energy consumption, equipment performance, resources utilized, asset & capacity management, environment compliances, fault diagnostic & predictive maintenance.

Through this, operators will be able to identify inefficiencies, optimize equipment placement, and plan for capacity expansion more effectively. Use of AI in DCIM will reduce maintenance cost and enhance lifetime through predictive analytics, improve data center environment and help data center operation staff.

The Future

As AI capabilities continue to evolve, the vision of a fully autonomous, self-healing data centre is no longer science fiction. Future AI-driven innovations could lead to:

● Self-optimising power grids that balance energy loads across multiple data centres in real time.

● AI-controlled microgrids that integrate renewable energy storage with dynamic demand response.

● Automated carbon tracking systems providing real-time emissions visibility for data centre operators.

The long-term goal is to leverage AI to manage the entire data centre lifecycle. This includes energy procurement, cooling management, predictive maintenance, and resource allocation, all with minimal human intervention.

Can AI save AI?

The AI revolution has the potential to thrive by building a sustainable and resilient digital infrastructure. AI is accelerating the growth of data centres, but it is also the most powerful tool to make them sustainable. The challenge for the industry is no longer whether AI can drive efficiency, but how quickly it can be implemented at scale.

With energy demand soaring, water resources under strain, and carbon reduction commitments growing stricter, data centres must embrace AI-driven sustainability as a core operational strategy and not as an afterthought. AI is either the problem or the solution. The choice is ours.

By Sudipta Sanyal, Principal consultant, Data centre, Aurionpro Solutions