The strategies behind data centre investment reflect overall best practice in investment, which is to balance risk against return in the context of specific requirements of time and outlay. There are certain idiosyncrasies in terms of data centres, in particular the fact that the cost of the equipment going into the data centre may be greater than the initial cost of land and construction, and the cost of operations (power, water, refit and refresh) will be greater than initial outlays. DatacenterDynamics (DCD) has calculated that less than 15% of the costs in a 20-year life will be incurred pre-commission (although an average figure cannot be easily calculated across the large number of possible variations). Investment decisions therefore need to focus as much on access to resources and the costs of these, and to establish a ‘whole of life’ view of the investment. An important aspect of this is the frequency of refit and refresh as the IT and network equipment housed by the data centre will evolve over time. As a rule of thumb, IT and networking equipment will update every three years and the infrastructure supporting it may need to also be updated.
The Business Case
As with any investment, a business case needs to be drawn up for the data centre based on demand for the services it will supply and the costs and the risk of delivering these. On this basis it may be more difficult to build a business case for either enterprise or for retail colocation than for cloud, IT services or wholesale colocation, since the key demand trends are towards the latter.
However, in established markets, the majority of large organisations will eventually operate a combination of their own ‘on-premise’ facilities, leased colocation facilities and a hybrid cloud platform to optimise their data infrastructure.
Since the global financial crisis, there has been an increased awareness of the need for risk management to drive decisions, and cost has become a greater factor, which means that the need to reduce risk is balanced against the need to constrain cost to a greater degree than was seen previously. At an enterprise level, this can result in lower levels of redundancy, reduced capacity or a lower-quality solution.
The first major factor in defining risk and return is the location where the data centre will be constructed. This needs to be looked at not over the short-term but throughout the life of the data centre. The focus of power availability has shifted from grid power to renewable energy sources and to guarantees about power pricing over time. The increasing primacy of cloud and networked data centres means a location that offers high levels of connectivity is an additional priority. It is also important to select a location that guarantees a quality of build and a quality of supply chain (vendors, consultants).
Where all of these factors coincide, data centre hubs are created – Singapore, Amsterdam, Hong Kong, Dublin, and a number of locations across Scandinavia are all examples of where the asset base is influenced by foreign inward investment, rather than local demand.
As with any form of construction-based investment, companies need to evaluate financial, operational, project execution, physical and regulatory risk. Risk management policies need to be aligned with compliance and regulatory policies, and should encompass technical and business continuity management and security. For issues such as data protection or compliance, the legal department will need to be involved.
Supply versus Demand
With investment decisions under increasing pressure and scrutiny, companies may be less willing to pay for building capacity that is not yet taken up. Modularity of build, and the construction of data centres in Lego-like units means that build can be synchronised with uptake. The process of matching revenue estimates against investment has to be revisited on a regular basis to take into account changing business conditions, particularly in a world where developments in software, cloud and miniaturisation may impact investments that are based on the leasing of space.
Forecasting is becoming ever more complex because the number of applications used are increasing exponentially as they are implemented and updated more regularly, which is proving to be more demanding of compute resources. For example, the aggregated power demand of the compute for cryptocurrencies is estimated to be already greater than the total energy consumption of Venezuela (placing it within the 40 largest electricity consumers globally). For this reason, ensuring security of supply for the data centre is a major investment priority, and colocation investors may spend as much as 20% of the overall investment on power coming into the data centre. Global cloud providers have already invested in renewable energy sources, as have major investors in data centres such as Goldman Sachs.
Supply/demand forecasting for a company with a very large data centre footprint will be easier than for a company with just one or two small DCs, because spare capacity will be easier to bring online. Forecasting must also take into account technological change and technology refresh, because new equipment is more powerful and able to manage higher compute loads without increasing footprint and potentially reducing power consumption. It must also take into consideration the different hosting models available today, or those that will possibly be available in the future, assessing what future workloads are likely to be hosted on a cloud platform or on SaaS.
Soft Risk Issues
Just as data centres and the IT world have moved on, so too have the requirements for investment. Location has to some extent been superseded by interconnection, the quality of facility has been balanced against access to quality service providers, and resource provision against the technology and expertise to deliver efficiency. Investment needs also to understand emerging factors such as data sovereignty, legislation on foreign business interests, or skills as a resource. As the data centre industry gets more adept at dealing with the resource issues that marked its legacy phase through better practice, more efficient technologies and through cloud computing, data sovereignty represents a new and less tangible challenge to the international free-flow of data.
Political stability, the ease of doing business, freedom from corruption and predictable taxation levels are also considered instrumental in reducing the risks of development and costs as well. The need to factor in a long-term view across all issues can be demonstrated by the costs faced to comply with the EU’s General Data Protection Regulation (GDPR) legislation and the considerable potential costs of non-compliance.
Levels of Return from Data Centres
Specific rates of expected return from data centres vary according to the factors identified through this article and also by the different measures. Most investors and investment advisers view data centres as offering above-average rates of return on investment when compared to other commercial property types. But the lower rates of return calculated for retail compared to wholesale, together with the higher rates expected for any form of colocation that will house cloud, indicate that the investment market is developing target niches as it matures.