In an era of rapid technological change, data centres now face increasing demands, with the rise of AI workloads.
The shift towards sustainability have led to the data centre industry turn to liquid cooling innovations as well as the integration of AI and machine learning for energy optimisation.
What are the emerging technologies shaping the future of data centre design and operation?
In a conversation with iTNews Asia, Siemens’ VP and Global Head of Data Centre Solutions, Ciaran Flanagan explores how new technologies, evolving business models, and closer collaboration with power utilities are driving the next generation of data centres, while also addressing critical concerns around energy consumption, cybersecurity, and regulatory compliance.
iTNews Asia: What emerging technologies do you believe will have the biggest impact on design and operation of data centres in your future?
Flanagan: We are at a big turning point in the industry, with huge changes ahead. The technical demands, especially with AI workloads, are increasing rapidly. But to answer your question about the key technologies impacting data centre design, I’d start by mentioning the shift from CPUs (central processing unit) to GPUs (graphics processing unit). This move is driving major changes in how we build and manage data centres. GPUs require more thermal management, power, and completely different layouts compared to traditional CPU setups.
At Siemens, we’re focusing on three key technology areas. First is liquid cooling. There are multiple liquid cooling technologies being used to handle the high heat from AI workloads, and our role is to provide the controls, sensors, and software to manage it. Liquid cooling is changing how data centres are designed and operated. It’s especially challenging to implement in older data centres, and there’s also a growing opportunity to reuse heat for things like district heating, vertical farming, or even converting it back to electricity.
Second, power consumption is a big challenge. AI workloads can cause significant swings in energy use, which means data centres need electrical systems that can handle these fluctuations. We’ll need to focus more on integrating renewables, improving power distribution, and using better software and telemetry to monitor energy usage.
Lastly, I’m really excited about digital twin technology. This technology lets us create virtual models of data centres to simulate and test scenarios—like heat buildup or system failures—before making real-world changes. It’s becoming an essential part of data centre management, and I think it will soon be at the heart of how data centres are operated. Digital twins are already used in design, construction (through Building Information Modeling), and simulation, and they’re helping teams plan, optimise, and reduce risks.
iTNews Asia: How do you envision the evolution of data centre design – will we see more modular or decentralised structures?
Flanagan: We’ll see more off-site, industrial construction of prefabricated and modular data centres. This approach will be driven by cost, time-to-market, and, most importantly, the ability to better manage risk and quality. The data centre will become more like a building block.
As for size, data centres will get both bigger and smaller. For large AI workloads, like with language models, we’ll need massive data centres. On the other hand, inference engines – especially at the edge will need smaller setups. This trend of having both massive data centres and smaller, edge locations will continue. Edge is a bit hard to define, but I consider anything under 2 megawatts to be edge though others might disagree.
I also think the data centre and power utility industries will converge more. While they’ll keep their separate identities, they’ll become more interdependent. Power is a major constraint for data centres, and the relationship with power utilities will become even closer.
– Ciaran Flanagan, VP and Global Head of Data Centre Solutions, Siemens
Networking technology is also maturing, giving data centres more flexibility in choosing locations. A few years ago, location was mainly about network access, but now power is just as, if not more, important.
We’ll see a big shift towards autonomous data centres. As data centres get larger and the industry faces challenges in hiring and retaining talent, more routine tasks will be automated, likely using machine learning and AI. This will be a major change in the next two to three years.
iTNews Asia: With the growth of new technologies, how is Siemens adapting its data centre strategies to accommodate decentralised data processing?
Flanagan: We take a multi-solution approach – sometimes we outsource to third parties for capacity, sometimes we use services from providers like SAP or Salesforce, and in other cases, we run our own data centres.
Why do we run our own? A few reasons: First, some of our data is sensitive and governed by local laws, so we need more control. Second, we have mission-critical data, like R&D or simulation data, that is crucial for product and technology development. Finally, some data is tied to long-term commitments with customers, like warranty or lifecycle requirements, so we store it in-house. However, many other services, like employee productivity tools (e.g., Microsoft), are outsourced to the cloud.
So, our strategy combines public cloud, co-location, and our own data centres for specific needs.
On hybrid and multi-cloud: Hybrid and multi-cloud architectures don’t present a huge design challenge in the short term, but they do when AI workloads come into play. For example, if you have a stable data centre and suddenly need to support AI workloads that demand 10x more power, energy, and cabling, managing that risk is key. Some customers use prefabricated containers or build new data centres for AI, while others keep AI workloads in their existing facilities. Digital twin technology can help here, by modeling the changes and managing risk.
When it comes to edge and hybrid cloud, there isn’t a one-size-fits-all approach. It really depends on the business. For example, a bank might centralise all computing, but a logistics or railway company will need a distributed cloud architecture to stay connected with assets spread across a region. Latency, connectivity, and business requirements drive the hybrid strategy, with AI introducing an additional risk to manage.
Automation is becoming essential as data centres grow in complexity. For example, we use AI to optimise cooling efficiency. Traditionally, data centres were over-cooled, which is costly. With AI cooling systems, we can operate closer to the thermal output, reducing costs. However, if conditions change quickly, human intervention may not be fast enough, so automation is necessary.
Machine learning can handle tasks that are too complex or time-sensitive for humans, especially when it comes to thermal and energy management, integrating renewables, and managing connections to local grids. Over time, these complex systems will rely on machine learning to run data centres more efficiently, and that’s where a software-driven approach becomes crucial.
iTNews Asia: What metrics do you think are most important for assessing the sustainability of data centre operations, biggest challenges in scaling data centre operations and how is Siemens addressing them?
Flanagan: In terms of sustainability, I believe the real focus in the data centre industry is on how we source energy. For example, if a data centre is mainly powered by coal, it doesn’t matter much if you use carbon-free concrete or have a recycling program for IT. The biggest environmental impact comes from the operations, particularly how you source your energy and water. That’s the real challenge for sustainability in data centres.
At Siemens, we offer tools and software to help measure and manage sustainability impacts, along with circular economy solutions and consulting for decarbonization..
As for how Siemens is addressing these challenges, we’re not in the power generation business, but we work with partners like Fluence and Siemens Energy to help customers access renewable energy. Our close relationships with the power utility sector are key to supporting data centres in navigating these challenges. From an operational perspective, we provide the software and control systems needed to integrate renewables without sacrificing reliability or availability. But ultimately, we rely on the power and renewable energy industries to provide the capacity needed for the data centre sector.
iTNews Asia: What are the biggest cybersecurity threats facing data centres today, and how can companies better prepare for these challenges?
Flanagan: Data centre systems like IT, cloud, and AI are exposed on the internet and constantly targeted for attacks. Every major enterprise worldwide is being tested and hacked all the time.
In the past, hackers would write scripts and manually try to break into systems, learning and improving over time. Now, they can use AI to deploy bots that sit on the network for hours, just waiting to see what happens. It’s a constant battle to protect the IT systems, and the same goes for the facilities side as well.
For instance, at Siemens, we work with major cyber security companies like Fortinet and track network activity in real time to help our customers stay protected. We also make sure our software evolves to address these threats as they come up.
It’s a constant challenge, but customers can feel assured that when they buy our systems, they’re buying something that’s been tested for cyber security, is hardened, and is continuously updated to deal with new threats. The threats are growing, but so should be our ability to manage them.
iTNews Asia: How do regulatory changes in data protection and privacy affect data centre operations, and what should companies do to stay compliant?
Flanagan: Regulators have come into the industry because they’re worried about its impact. If we look at different regions, the European Union is being very proactive. They’re requiring data centres of a certain size to report on energy use, carbon impact, and sustainability efforts. We’re starting to see similar moves from the EPA in North America, China, India, and Singapore. These governments have already taken steps to manage data centre growth due to sustainability concerns.
The regulatory framework is here, but the challenge is that some of the regulations are being made by people who don’t fully understand the industry. They’re more reactive to issues like energy consumption and sustainability rather than being designed to improve efficiency and performance.
For operators, this means they need to be better at gathering data on their energy consumption, water use, carbon footprint, and overall sustainability impact. They’ll need to show they’re actively managing their impact. The goal will be to measure, manage, and mitigate the impact. Governments may incentivise operators to provide on-site energy generation, district heating, or even to integrate with the local energy grid.
We’re still in the early stages of regulation, and right now, it’s mostly about energy use and sustainability. In the future, regulations might expand to cover water, waste, and recycling, but for now, it’s all about managing the energy footprint.
iTNews Asia: How do you think new business models, like “data centre as a service,” will shape the future of data centres, particularly in terms of energy efficiency and sustainability?
Flanagan: The industry has definitely matured, and running data centres efficiently now requires a lot of expertise. Many businesses don’t develop that expertise because data centres aren’t their core business. As we move more workloads into larger, scalable facilities, efficiency becomes even more important, and that’s a big part of the future.
Looking at recent trends, like AI, I believe that services like GPU-as-a-service will require specialised expertise, and that’s a good thing for efficiency. So, while companies like Siemens will still need some in-house data centre capacity for specific needs, the major workloads will likely be handled by experts. These experts are continuously investing in improving their operations, which will drive greater efficiency. In the future, platforms like GPUs, public cloud, and GPU-as-a-service will be key to achieving even more efficiency, and that trend will continue.