‘Hybrid cooling is now a pragmatic default strategy’

Given the surge in data centre activity and the sheer scale and ambition of projects, what design planning and execution solutions are you providing to support an integrated project delivery approach involving civil, structural and MEP disciplines to improve reliability and energy efficiency of data centres?

‘Surge’ is really an understatement right now. We are seeing an absolute boom driven heavily by AI and cloud. And when you are building at hyperscale level, working in silos simply does not work anymore.

At Meinhardt, our approach is all about integration from day one. We rely heavily on advanced BIM modelling and digital twins, so our civil, structural, MEP and architecture teams are effectively in the same virtual room before any construction activity has even started. That allows us to identify and resolve clashes early.

But it is not only about avoiding clashes. It is also about optimising the whole system. By integrating the design early, we can optimise the building orientation and the structural layout to support highly efficient airflow and cooling distribution. We are also pushing hard into modular and prefabricated design. By building standardised, highly efficient MEP components off site, we can accelerate the construction timeline while ensuring the final facility is reliable and energy efficient. Ultimately, it is about using technology to make complexity feel seamless and to reduce capital expenditure to the lowest possible value.

Are you able to quantify the improvement in energy efficiency through this approach?

Yes. The climate conditions in the GCC region are, of course, challenging, but we are seeing clear gains in power efficiency. Previously, in markets such as Saudi Arabia and the UAE, we were achieving PUE values of around 1.5 – 1.55. Now, that is moving towards 1.4 and below, which is a strong achievement.

Air cooling systems address the cooling needs of about 80% of data centre cooling projects around the world. Are you seeing any shift in this? For instance, is there a pivot happening towards immersion cooling or direct-to-chip cooling? And if so, are these positively impacting energy efficiency and helping lower greenhouse gas emissions?

Yes, a massive shift is underway. Air cooling has been the workhorse of the industry for decades. But with AI workloads pushing rack densities from a standard 10 or 15 kilowatts up to 50 or 80 kilowatts, and sometimes even more than 100 kilowatts, it simply cannot move enough thermal energy fast enough. So, we have to rely on liquid cooling.

We are absolutely seeing a strong pivot towards liquid cooling, specifically direct-to-chip cooling and immersion cooling, because it targets the heat right at the source, whether it is the CPU or the GPU. Immersion cooling, where servers are submerged in fluid, is also highly effective for certain thermal management requirements.

And to your point about efficiency and emissions, yes, they make a huge positive impact. Liquid is more efficient at transferring heat than air. By shifting the heavy thermal lifting to liquid, it reduces the power needed to run massive fans and chillers.

What is your view on hybrid architectures that combine air cooling for space-level requirements and liquid cooling for dense loads in new-build as well as brownfield projects?

I am really glad you brought this up because hybrid cooling is now a pragmatic default strategy. Think about it. A facility might have an AI cluster running at a 100-kilowatt rack, but it may also have standard networking and storage equipment running at 10 kilowatts.

And sometimes, when we are talking about building the AI model, you need the high-density racks. But later, when you start running the model, you do not actually need the same density or the same capacity. So, it does not make financial sense to put everything in a liquid bath.

In a hybrid model, we use liquid cooling specifically for the high-density AI loads. And we use optimised air cooling with hot and cold aisle containment to manage ambient room temperature and lower-density loads. For brownfield or retrofit projects, this is an absolute game changer. You do not have to rip out your entire legacy HVAC system. You simply bring liquid cooling to the specific rows that need it, which saves a massive amount of capital expenditure and downtime. For new builds, designing shared infrastructure that supports both air and liquid from day one gives the operator incredible flexibility.

Geopolitical events are causing supply chain uncertainty in the case of critical equipment, including cooling infrastructure, on a global scale. How should the world plan for fast-developing situations?

It is one of the most pressing challenges we face right now. You can have the best design in the world, but if a critical chiller or generator is stuck on a ship or delayed at the factory, the facility cannot open.

To plan for this, we have to fundamentally change how we procure. We can no longer rely on just-in-time delivery. We advise our clients to commit capital early to secure manufacturing slots, either by direct procurement or by having framework agreements with suppliers. In some cases, critical equipment is ordered almost immediately, sometimes before the final design is fully locked, or immediately after the concept design stage is finalised.

At the same time, we cannot rely on a single supplier or a single geographic region. The supply chain has to be diversified. We have all seen how quickly situations can change. One day, you may find the Suez Canal blocked. Another day, you may find Hormuz blocked. So, you need to diversify delivery locations, as well. In some instances, we may accept a slight premium to use secondary suppliers or to source equipment locally or regionally, simply to guarantee delivery.

And in our financial and project models, we always include schedule contingencies, essentially budgeting for temporary solutions such as rental generators or temporary cooling, in case permanent equipment is delayed. It is all about pricing in the cost of risk and building extreme flexibility into the project plan.