Global demand for data centres is booming. With use of artificial intelligence and cloud computing surging, applications and data need to be housed securely and, to provide high-speed access, near to customers.

At Arup, our clients are planning new hyperscale and edge data centres across the Asia-Pacific region to meet rising demand, diversify their base and improve latency (the speed of data movements). They are among a host of global data centre operators and big tech companies searching for suitable locations.  

To speed up the development process and secure economies of scale, many data centre companies use standard reference designs – creating, in an ideal scenario, a series of almost-identical data centres around the world. Smarter owners and operators adapt their reference designs and programme to each location, investing more time in the design and planning phase to achieve a shorter construction phase and better outcomes.  

So, what needs to be adapted to suit local conditions, and why? 

Location, location, location

Imagine you’ve already built twenty data centres in the UK, and you’re planning another. You can probably simply replicate your existing design, incorporating everything you’ve learnt so far.  

But what if you want to move into a new market – Thailand, for example? There are obvious physical differences: climate, topography, risks of natural disasters such as flooding or earthquakes. There are regulatory differences in planning, fire regulations and laws. And there are different utility networks to interface with: power, fibre IT and water – each with its own constraints and requirements. Data centre companies do carry out standard due diligence checks and research before acquiring a site. But if this diligence isn’t detailed enough and specific to the location then procurement and construction may not go smoothly.  

What materials are you planning to use for the building? Steel is usually quicker than concrete, but do local contractors undertake steel construction regularly, and can they do so at the pace set out in your schedule?  

If you have a global contract with an equipment supplier, will their equipment still fit neatly into the available space after you’ve adapted the layout for local fire regulations? Is that equipment approved and certified in that location? If it can be used, will additional authority reviews and approval processes affect the schedule? Is there enough skilled labour locally to install and maintain that equipment, and how long will it take to get everything signed off? Virtually every data centre project will need to make some adaptations to the construction process and programme in a new location.  

Even small delays, or reductions in a data centre’s targeted IT delivery capacity, will translate into increased costs and lower long-term profitability. If the schedule and the capacity do remain the same, something else may suffer instead: the quality of construction, the efficiency of the design for the local context, or how easy the data centre is to maintain or adapt. Any of these could also affect the bottom line. 

Data centres: one global template doesn't fit all illustration
Data centres: one global template doesn't fit all illustration

Assess and reduce risk early

Carrying out early-stage assessments in each new location reduces risk and prevents unwelcome surprises in the latter stages of the project. These assessments should include not only typical site due diligence and a high-level data centre test fit, but also the suitability of the ideal reference design and schedule to the local context. A thorough up-front assessment typically results in a faster, smoother, cheaper construction process – without expensive late design changes or compromising on key metrics.  

Consider the Asia-Pacific region, for example. It’s an atypical market undergoing rapid growth. There is a wide variation in codes and regulations across the region, different languages and cultures, different degrees of maturity in the data centre construction industry, as well as a combination of developed and developing countries with their own methodologies and norms. A design and programme that worked successfully in the UK or the US is quite likely to need substantial adaptions to work in Tokyo and further adaptations to work in Thailand, Indonesia, and so on. But those learnings will then be applicable to subsequent data centres nearby in similar markets. Understanding those similarities in the market and planning correctly first time streamlines the second project in that location, and so on. The third, fourth and subsequent data centres will be far quicker and easier to develop – leading to an overall market advantage. 

Time invested in adapting plans for a data centre to suit the local context is always worthwhile, and can identify opportunities, as well as risks.

Nicholas Long

Associate Director

To future-proof data centres, look further ahead

In a fast-moving industry, hurrying to meet present demand risks forgetting to consider the future. Around the world, governments are moving to regulate and incentivise data centre operators – who are significant consumers of power – to work towards net zero through measures such as carbon pricing.

For most data centres, on-site generation is unlikely to be significant, so buying or investing in renewable energy from local utility companies will be one of a set of important decarbonisation strategies. With net zero deadlines fast approaching, it therefore makes sense as part of a wider sustainable development strategy, to explore whether this is feasible in new locations and estimate the carbon cost of a data centre’s power needs alongside the financial cost.

Resilience to the effects of climate change is increasingly important, and we encourage our clients to go beyond minimum requirements by considering, planning and implementing sustainable development. Considering early how to integrate sustainable development into a data centre offsets operational and financial risks to the site, enabling the full-life value of the facility to be realised.

For example, it became clear following the severe flooding South Korea experienced in 2022 and 2023 that government hydraulic flood-risk mapping, often used as the basis for designing buildings, had not kept up with the latest rainfall statistics, let alone future predictions. So protecting a data centre against flooding for the next 20 years would mean going beyond current minimum requirements, quantifying the likely flood risk in the prospective location using forward-looking projections and designing in reasonable safeguards to protect the development’s value. Sustainable development is thus not only about doing the responsible thing for the environment and society at large, but also helping to enhance and retain a development’s investment and long-term value. 

Climate change also threatens data centres’ operations. And this too depends on the local context. In water-scare areas it could become harder to access to water to cool the data centre, or local utilities who rely on that water for hydroelectric power may impose constraints. Where temperatures are rising quickly, equipment installed today may soon become undersized or space allocated for these systems could be insufficient to add more plant. Assessing the data centre in its local context, both now and throughout the anticipated life of the facility, is essential to its longevity and to help ensure the facility can remain operational and profitable. 

A reference design is a useful tool, but there really is no one-size-fits-all solution, so developing a strategy to localise as a pre-development plan is the best way to achieve strong returns now, next year and in the long-term.