The Evolving Conundrum of Site Selection

Location and technical assessments are key, but other factors also come into play

In recent years, I have had the privilege to speak with a handful of investors and operators, and one question that often comes up is: “What makes a site a good choice for data centre builds?”

I have two default answers when asked difficult questions in public: “It depends”, and “the devil is in the details”. But behind closed doors, I am more candid and I collated some points from conversations with a business associate who is looking for a new site for her clients who were driven out from a certain geography.

It is all about the money

Let’s go back to first principles and agree that all businesses are about making a profit. The site selection is part of a series of business decisions that influences the profitability of the investment. The same has been echoed many times by Art Carapola in his book ‘The Data Center Builder’s Bible Book 2: Site Identification and Selection’ on site selection, though there are some exceptions to the rule.

Checklists do not replace oversight and discretion

If you take the same question to the brokers and due diligence providers, they will rely on their experience and may or may not take reference from the plethora of technical checklists and scorecards. However, a handful of consultants are blinded to the fact that while checklists are very useful, they alone may not adequately articulate each organization’s complex and unique business needs and risk appetites. One consultant tried to lie either to himself or his client in Singapore that there are no waterways near the recommended site, because that was the only way to get the desired score under the site selection checklist.

Similarly, a particular customer was blinded to the fact that they chose a site that was flood prone. It was easy for them to get a copy of the map that shows which areas of the country were at risk of flooding, but they chose not to. They decided the cheapest site was the best site and ignored any facts that might have challenged that rationale and have since regretted their decision.

Confluence of geographical advantages

There are some obvious geographical advantages to certain sites, such as high elevation, proximity to high tension power, or high density of dark fibre networks. There are also clear advantages in being far away from certain locales such as airports, factories, major roads, embassies and military installations so as to reduce the risk of certain threat scenarios happening.

In some cases, the sites are deemed attractive enough to overlook other constraints. That could be the rationale for Microsoft to build its own power station to power its sites at Grange Castle Business Park in Dublin, Ireland.

Joshua Au, Singapore Chapter Lead, Infrastructure Masons

Some disadvantages do not sink the boat

Despite the fact that an area was flood prone, it did not stop a particular operator from building a modestly profitable facility in a particular country in APAC. It has since built several more facilities in the country which are Tier 2, again showing that not all sites need to be built to Tier III/IV or Tier 3/4 specifications to make business sense.

Though there are clear technical disadvantages to building a data centre several floors up in an office building in a bustling city, that has not stopped paying clients from doing so in metros such as Tokyo and London.

There are several sites at the outskirts of Shanghai and Beijing, which are more than 2 hours’ drive away. To some customers—the likes of gaming companies—the travel distance is not a deal-breaker.

In other cases, the technical assessment is evident. There is still inadequate power generation in certain parts of Asia, and a lack of human talent needed to design, build, and operate a site. The skills needed to build a data centre are vastly different from the skills needed to build hotels.

Counting the cost

Each data centre build is a marriage between the data centre operator and the local economy (and authorities) for the life span of the data centre, and it is in the interests of the operator to be intimately familiar with the compliance regime. I vaguely recall more than 40 permits are needed to build in India; in Singapore at least 20 permits are needed and operating costs such as taxation, regulatory and utility costs.

Data centre builders are aware that not all sites and local economies are created equal, and that the returns on investment, and the corresponding risks and time frames differ from country to country. Suffice to say, there is a subtle difference between an underserved market and an unprofitable one.

What is good for the goose may not be good for the gander

The guidance notes for the valuation of data centres by the Royal Institute of Chartered Surveyors, UK talk about the varying attitudes of local planning authorities to data centre development proposals.

Each data centre is a centre of gravity that changes the layout and geography of the Internet, the behavior of Internet traffic, and improves the competitiveness of the connected digital economies.

Hence, there are some policy makers who are more prepared than others to put up certain incentives to make their county or country a favorable choice for data centre site selection.

Some policy makers are also more intimately attuned to the needs and pain points of data centre operators. Norway’s data centre strategy paper comes to mind, including its deliberation on property tax. Some governments are increasingly cognizant that the interconnected systems of data centres, subsea and terrestrial routes serve as the global supply chain of digital data and services and would seek to influence the geography of the Internet based on their political  prerogatives, and there have been data centre and subsea cable proposals that have suffered when political winds blew in a different direction.

There are also the interests of local stakeholders, communities and supply chain vendors to contend with. Some operators find out too late that the change in political winds and the corresponding regulatory cost and uncertainty make profit an elusive goal.

Perceptions influence decisions

It is often about pandering to the customer in mind because perceptions change the sentiments and purchasing choices of policy makers, operators, and consumers.

What is acceptable to a hyper-converged site may not be so to a traditional colocation provider. A facility supporting a bank would have very different needs from a facility meant solely for cryptominers. The assessment of a site’s merits is an expertise based on continually reflecting on the lessons from past assessments, the ones that went right and especially the ones that went wrong. This involves not only technical assessment, but also business acumen to make shrewd trade-offs between competing priorities and considerations. Because success is not about a data centre being built to a set of technical specifications, but how the data centre contributes to business outcomes throughout its lifetime. Reframing the facts to a new narrative to an interested investor or customer could sometimes be all it takes to make a site a good choice. Is a data centre built underground a good idea? What about a secure converted underground nuclear bunker data centre drawing glacial water from a deep subterranean lake, eliminating the need to pay for and power expensive chillers? It depends more on the customer in mind than the answers on a checklist. I would be glad to hear your thoughts on the above and exchange stories.