How digital twins are changing the way data centres are designed

An increasing number of Australian organisations are undertaking digital twin initiatives; however, far fewer are using the approach when developing new data centres. This needs to change, as digital twins have the potential to rewrite the design rules for data centres and drive improved decision-making across every business function.

The digital twin

Essentially, digital twins are virtual copies of physical assets that can be used to replicate entire data centres or select subsystems within them. They can be used to forecast almost any scenario, from identifying opportunities for greater efficiency to disaster simulation and infrastructure planning.

By leveraging this capability, operators can better forecast at the development stage how new projects are likely to impact local communities and identify locations best suited to deal with the demands of heavy AI workloads.

This capability is becoming increasingly important as forecasts¹ predict that AI racks could consume between 20 and 30 times the energy of traditional racks by 2030. As well as supporting plans for construction, digital twins can make recommendations to balance existing workloads and identify areas for greater efficiencies in real time. For instance, suggesting re-configurations for server racks rather than building new physical racks.

Barriers to effective deployment

It must be remembered, however, that the value derived from digital twins is significantly reduced if they are not supported by a solid ‘lake’ of data which draws from connected information technology and operational technology (OT) systems.

Unfortunately, many data centre operators are still dealing with data that is dispersed across IT/OT environments. For example, unless OT systems such as a power distribution system (PDS), uninterruptible power supply (UPS) and generators are integrated with IT systems, digital twins will provide operators with inaccurate forecasting. This situation could lead to scenarios where systems are overloaded or deprovisioned during peak demand.

Accurate disaster simulations are equally dependent on IT/OT convergence. Take the example of a data centre provider simulating a regional power grid failure. Primary power is cut, back-up generators take over seamlessly, and failover scripts reroute workloads to a secondary facility.

However, heating ventilation and air conditioning (HVAC) and building management systems (BMS), which are not integrated with IT systems, continue running in standard operating mode. The HVAC system, functioning separately to the simulated power outage and increased IT load, does not adjust cooling to match the spike in server activity.

In a worst-case scenario, this could result in server aisles nearing thermal shutdown. When systems are not ‘in dialogue’ with one another, and the physical interdependencies between IT and OT systems are not well understood, operators end up with a false sense of readiness.

Invest now for the future

Digital twin technologies require a significant upfront investment for businesses. Real-time data collection requires the installation of hundreds, sometimes thousands, of IoT sensors across physical assets to capture operational data.

There is also an upfront cost where newer systems require integration with legacy infrastructure, and processes such as data cleaning and normalisation are required.

While long-term benefits such as predictive maintenance, process optimisation and operational resilience are well recognised, the time required to realise these gains often means businesses are hesitant to take the leap and invest. Many businesses are also still figuring out how to accurately measure and benchmark ROI for these investments.

It should be remembered that digital twins can also play a crucial role in disaster prevention. By simulating emergency scenarios, such as the impact of extreme weather events on power supply or equipment failures, predictive algorithms enable operators to proactively identify vulnerabilities and implement safeguards before issues arise. This in turn can significantly reduce the risk of unexpected outages and minimise the financial impact associated with downtime. In fact, research shows that the Global 2000 companies lose $400 billion annually due to unplanned downtime.²

Another area where digital twins are already making a difference is in defining and delivering more precise service-level agreements (SLAs), especially as the industry shifts towards software-as-a-service models. For example, digital twin technologies can be utilised during co-creation processes with end users to guarantee that a solution designed by the service provider is in line with expectations. The value delivered by digital twins will continue to grow as further uses are identified. By investing in technology today, organisations can be well placed to take advantage of these benefits in the future.

<sup>1. Techradar 2025, This graph alone shows how global AI power consumption is getting out of hand very quickly - and it's not just about hyperscalers or OpenAI, <<https://www.techradar.com/pro/security/this-graph-alone-shows-how-global-ai-power-consumption-is-getting-out-of-hand-very-quickly-and-its-not-just-about-hyperscalers-or-openai>>
2. Oxford Economics 2024, The hidden costs of downtime: The $400B problem facing the Global 2000, <<https://www.oxfordeconomics.com/resource/the-hidden-costs-of-downtime-the-400b-problem-facing-the-global-2000/>></sup>

Top image credit: iStock.com/monsitj