Mark Fenton, Product Engineering Director, Cadence, says that to accommodate AI’s growing demands, facility leaders should prioritize optimizing what they already have.
AI’s rapid growth is driving soaring power demands in data centers, the foundations of the technology’s capabilities.
Now, Europe’s data center energy consumption is set to surge from 62 terawatt hours (TWh) to over 150TWh by 2030, largely due to AI advancements.
Yet, many organizations are failing to track the environmental impact of AI effectively. This is where data centers can lead the charge, leveraging digital twins to align AI ambitions with sustainability. These intelligent virtual replicas of the data center optimize power management, improve energy reporting, and allow for improved cooling systems – all of which are crucial steps to meet AI’s GPU-intensive demands while reducing carbon emissions.
AI has some phenomenal use cases. For example, the 2024 Nobel Prize in Chemistry was awarded to DeepMind for its work on an AI system that accurately predicts protein structures, solving one of biology’s fundamental problems. Further to this, AI has been used by police departments to identify at-risk youths, and it’s being used to help with future tsunami planning. Overall, its possibilities are endless and lifesaving. However, its impact on the environment and its strain on data centers can no longer be an afterthought.
The University of California, Riverside found that OpenAI’s GPT-4 model consumes the equivalent of three water bottles to generate just 100 words. As a result, data centers could use a total of 1,000TWh annually by 2026, roughly equivalent to the electricity consumption of Japan, says the International Energy Agency (IEA).
To put this into perspective, Japan has a population of 125 million people. Though AI growth and sustainability seem counteractive, the two can go hand in hand.
Harnessing AI to Make Data Centers More Sustainable
AI’s rapid advancement is clearly placing an increasing strain on data centers and the environment. However, the very technology driving this demand – AI – can also be part of the solution.
Digital twins, when enhanced with AI, offer a smarter way to tackle sustainability challenges, allowing operators to improve operations within their facilities without disrupting daily operations.
A digital twin is a virtual replica of a physical data center, enabling operators to simulate scenarios, test solutions and fine-tune processes before making real-world changes. When paired with AI, digital twins provide predictive insights, analytics, and automation that help data centers make evidence-based decisions to become more energy-efficient. There are many ways in which data centers can utilize the capabilities of digital twins and AI, from reducing overprovisioning to optimizing cooling and enhancing energy reporting, all of which can significantly drive down energy use.
Eliminating Overprovisioning to Reduce Energy Waste
Firstly, digital twins can drastically reduce the unsustainable practice of overprovisioning, a practice where facilities deliberately allocate more resources than necessary to ensure peak demand can be met safely. While overprovisioning ensures continuity, it also leads to significant energy waste.
Cadence reports that enterprise data centers can experience stranded capacity losses exceeding 40%, primarily due to over-provisioning during design.
By introducing digital twins, data center operators can gain valuable insights into resource utilization and optimize capacity planning much more accurately to significantly reduce overprovisioning. In turn, this reduces unnecessary energy consumption and creates a more sustainable approach to capacity management.
Optimizing Cooling Systems for AI’s Energy Demands
Another key tactic to drive down energy usage is to explore how data centers are being cooled. Cooling is one of the most energy-intensive aspects of data center operations, especially as AI workloads drive up power consumption. Digital twins, however, make it more attainable for data centers to benefit from liquid cooling. A fundamental reason for turning to liquid cooling is that high-density server racks, demanding workloads, and steadily rising power densities are pushing the limits of air cooling. Air cooling can readily handle heat loads of up to around 20kW per rack. However, beyond 20-25kW, a combination of direct liquid cooling and precision air cooling becomes more efficient and economical.
By adopting liquid cooling using digital twins, data center operators can uncover what they typically can’t readily see or measure – including cooling efficiency. With this technology, operators can evaluate the potential benefits and drawbacks of different liquid cooling methods before making any physical changes. Digital twins also allow operators to test various scenarios, such as analysing how and where to introduce liquid cooling into an air-cooled data center. The result is a tailored solution that meets specific heat load requirements.
Enhancing Carbon Reporting and Compliance
Elsewhere, digital twins support the sustainability journey of data centers through energy reporting. With new regulations such as the Climate Change Agreement and the EU’s Energy Efficiency Directive (coming into effect in May 2025), accurate carbon emission reporting is more critical than ever. Digital twins automate emissions tracking, giving data center operators precise, real-time insights into energy usage, inefficiencies, and areas for improvement.
Detailed reporting not only helps to meet reporting standards but enables data centers to provide management with a comprehensive understanding of potential energy savings and efficiency gains. For example, the technology can identify inefficiencies caused by stranded capacity, which refers to unused or underutilized power, cooling or space resources that have been provisioned but are not actively supporting IT workloads. This data-driven approach ensures facility leaders can report their findings accurately and use the same information to make informed decisions that prioritize energy efficiency and reduce environmental impact.
Accelerating the Transition to Renewable Energy
Beyond improving efficiencies, eventually, data centers will be able to turn to digital twins to accelerate the adoption of clean energy. While the industry looks to harness more renewable power to ease grid and environmental pressure, these solutions are complex and require meticulous planning. Data centers need a testbed before implementing drastic changes.
That’s where facilities will eventually begin to leverage digital twins to test the potential impact that clean energy methods, such as wind turbines, solar panels, smart grids and hydropower systems, have on operations. In theory, this will allow data center operators to visualize how renewables integrate before investing in new infrastructure. For example, they can determine the most efficient configurations for their specific needs by simulating different cooling strategies and resource allocation alongside renewable energy fluctuations. Ultimately, this approach will ensure data centers get the most out of their renewable energy sources while minimizing disruption.
Impactful Outcomes
Data centers are at an inflection point: supporting the use of AI whilst meeting sustainability measures under stricter regulations. By using digital twins, impactful environmental outcomes will be second to none. While there is a way to go in terms of reducing the energy consumption of AI, data centers must take control of their carbon footprint. Rather than simply expanding capacity to accommodate AI’s growing demands, facility leaders should prioritize optimizing what they already have. By embracing digital twins, data centers can enhance efficiency, cut emissions and integrate clean energy sources. This is the only way to ensure AI can reach its full potential, sustainably.
