Did you know that over 90% of the world’s data was created in the last 18 months to two years? That’s all electronic data ever created. The demand for data storage is growing at a rapid rate as the cloud computing megatrend continues. Covid-19 has accelerated this demand with people continuing to work from home post lockdowns.
Technology business models have evolved significantly over the last two decades. Increasing internet connectivity speeds have allowed computing resources to be aggregated in the “cloud” rather than at the user’s premise. Technology services are now being delivered at a lower cost and “as a service” at an increasing rate. Everything from software applications, streaming entertainment and backend databases are moving to the cloud. Computing resources are cheaper at scale compared to at a single user’s premises. With increasing connectivity in our daily lives (fixed wireless and 5G) this trend is unlikely to end anytime soon.
What are Data Centres?
Any organisation that uses or creates data has the need for a data centre (aka file server) at some point in time. Traditionally these data centres were on premises and the organisation’s internal IT infrastructure team would maintain them. With the amount of data being produced increasing significantly and continuously changing dynamics of the IT ecosystems, organisations have moved their data centres to large scale co-location, cloud, enterprise and managed service data centres. These are the modern-day data centres we hear about in the news. On the outset, they may seem complex structures, but put simply, they are a physical facility leased by multiple customers to house their data.
Data centres offer their equipment to customers who don’t want to, or cannot, invest in setting up the same facility for private use. The client pays based on consumption and has the added advantage of being able to scale up quickly if their demands increase. Data centre operators determine fees based not only on the square footage each customer uses but also the amount of electricity used or provisioned. Electricity usage is a critical cost driver, both from the direct use by servers, but also because every KWh of power consumed translates into waste heat and, hence, an air-conditioning requirement.
The datacentre service provider will be responsible for the maintenance and upgrade of all the equipment. Also providing other services including access to broadband, interconnectivity with other customers, security and onsite professional services. In other words, data centres offer the combination of interconnection with high-speed enterprise access to data and an ecosystem that will enable scalable and agile infrastructures, bringing the customer and its applications to the network.
There is huge growth potential for the industry, in Australia alone, revenues are projected to grow from $4.7 billion in 2021 to $9.1 billion by 2027 – that’s close to a 100% growth in revenue in six years.
Infrastructure, Property or IT……or all three?
One of the key characteristics of data centres that sets them apart from their real estate peers is the higher demand for, and emphasis on, power — both the immediate need to run and cool thousands of servers as well as the backup power on hand that can seamlessly keep those servers running for up to a week or more, should the primary power source go down. To give some perspective, on a global scale, data centre power consumption amounted to about 416 terawatts, or roughly 3 percent of all electricity generated on the planet.
When considering the real estate aspect, well we all know how the saying goes in property investing… “location, location, location”. Data centre success and occupancy has been correlated with the location of the data centre. Data centres close to the operations of their customers are preferred due to connectivity speed as well as the ease of access by the internal IT team. Bringing data centres close to consumers also speeds response times (aka latency). As a result, data centres in major business hubs have seen higher occupancy rates as well as better margins.
In order to achieve those margins, there are two main profitability factors to consider, controlling costs and achieving scale. Large companies benefit from significantly reduced marginal costs, with fixed costs shared across a greater number of customers. Additionally, the number of employees required to manage each server declines as the size of a data centre increases. These economies of scale favour larger companies, increasing the industry's market share concentration. Change in the industry is quite rapid, major companies have been re-investing profits to keep up with the highest speed servers and network architecture, as well as employing techniques that enhance energy efficiency and reduce the carbon footprint of their data centres.
At an individual project level, careful consideration needs to be given to the forward-looking EBITDA projections. Simplistically, a good infrastructure project has high upfront capital expenditure, some form of ‘moat’ (monopolistic style barriers to entry) stable revenues, and high operating margins throughout the life of the project. Datacentres can definitely deliver attractive operating margins. For example, Next DC delivered an FY21 operating margin of 55%.
The revenue profile of a data centre consists of varying contract lengths. However, companies focus on securing long-term credit-worthy clients to help to help boost their own credit profile. Looking at Next DC, we observed that 43% of the Next DC’s revenue was from two clients (weighted average contract length unfortunately was not revealed). This adds another level of complexity in valuing data centre projects as the differing contract lengths of customers, their credit profile, and the contractual terms are key inputs to finding the correct valuation for the project or level of debt each project can sustain.
Similarly, the operational complexities of continued technological investments and expansions, makes a data centre less like an infrastructure project (where generically the bulk of capex occurs at the start of the project life) as there is a constant need for continuing capex to maintain relevance and competitiveness in the market – again Next DC in the infographic provides an example of the continuing capex spend.
In summing up, data centres present some unique features and risks for investors / debt financiers compared with other asset classes. Whilst they share many characteristics with traditional infrastructure, they also have a number of non-infrastructure characteristics. We believe data centres are still in the early stages of an emerging sector, valuation models are expected to evolve and mature with time as they factor in the multi-faceted nature of this asset class.