AI-driven data demands challenge sustainable data center operations but solutions like digital twinning can ensure efficiency, safety, and resilience
Back in 2018, the world generated 33 zettabytes of data. By 2020, this number had nearly doubled to 60 zettabytes, soared to 120 zettabytes by 2023, and has been projected to hit 180 zettabytes by 2025.
IDC predicts that by 2028, 80% of CIOs will leverage organizational changes to harness AI, automation, and analytics to drive agile, insight-driven digital businesses. Data generation will continue to increase as AI adoption rises.
Today, a single ChatGPT query consumes 2.9 watt-hours of electricity compared to a browser search’s 0.3 watt-hours. With the escalating usage of AI technologies, these data and power consumption figures will only climb.
The question therein lies: Is increasing demand for data centers jeopardizing sustainability? How can we manage the AI boom that is here to stay?
Powering DCs will be 10x harder
Truth be told, data centers (DCs) are consuming a lot of energy. They currently account for 1 to 1.5% of global energy consumption and this number will increase significantly.
Responding to this transformation is going to take different tools, technologies, and ecosystems that exist today. There are also promising new technologies, but the real challenge is going to be driving the demand curve forward and doing it sustainably.
As DC designers, builders, and operators, we will face a 10-fold increase in technical challenges in the coming years, grappling with 10 times the amount of cabling, structure, networking, power density, heat loss and complexity issues. We need to brace ourselves. A traditional data center model will struggle to keep pace with this. For instance:
- AI workloads are known to cause significant fluctuations in power usage and require adaptive power management systems that can dynamically adjust to varying load demands to mitigate power swings.
- The increasing DC energy density amplifies the risk of operational failures. Hence, implementing rigorous monitoring and real-time anomaly detection systems becomes crucial to minimize errors. The DC industry has approximately 15 to 30 minutes to deal with thermal runaway before catastrophe, but with AI-linked power swings and increased workloads, we are probably looking at just a few minutes during a cooling failure before catastrophic thermal effects occur.
- Also, the elevated power demands associated with AI can heighten fire risks. This will require operators to install advanced fire suppression systems tailored to tackle high-intensity fires. Additionally, they must ensure that their data center design accounts for the unique needs of high-powered AI racks.
- And AI advances, cyber threats will become more sophisticated, and hackers and bad actors will use similar technology to exploit vulnerabilities and launch more targeted attacks at DCs.
The path to a sustainable data center industry involves sourcing the right technology, utilizing decarbonized energy, and minimizing water consumption, and then applying these resources wisely through energy efficiency, grid participation, and heat reuse.
In modern DC operations it is crucial to balance these elements effectively to ensure efficiency, safety and security.
Using a digital twinning solution
Acknowledging the complexity behind addressing these needs, DC operators must consider tailored solutions that not only address market demands but also the imminent power adjustments needed with AI workloads.
Having a digital twin of a data center can help accomplish this. Digital twins manage and mitigate the risks that emerge as we embrace this new age of supporting AI infrastructure.
- It gives designers the ability to model the facility in design state, allowing us to understand the interplay between the different infrastructure elements
- It also provides extensive simulation capabilities, where one can simulate failures from thermal management; massive increases in electrical workload; the impact those increases will have on existing electrical infrastructure and possibly on the broader grid network to which the data center is connected
- The technology not only mitigates the environmental impacts but also elevates the economic dynamics of digitalization to support a future where technology and sustainability are intertwined
One new DC in Estonia was named the most sustainable data center in the Baltics when the use of digital twinning allowed the operator to rely solely on renewable energy for power. Smart technology allows the DC to increase efficiency and leverage the surrounding environment to advantage, illustrating the attainability of constructing DCs embodying reliability, security, and sustainability.
In the same way, DCs in the Asia Pacific region have the same potential to innovate in the way they are built and operated. We should not be afraid of the challenges. With the right partnerships, tools, and approaches, we will be able to manage the increasing demand, and we will be ready to take on the challenge of AI.
Just as every watt saved contributes to the broader goal of energy efficiency and sustainability, each step taken by DC operators towards smarter, more sustainable operations will play a crucial role in shaping a more ecologically responsible industry.
