$229 billion. That was the estimated size of the global data centre market in 2023. And with an estimated annual growth rate of just under 10% we could be looking at a data centre market in excess of $400 billion by 2030. The UK already hosts the highest number of data centres in Western Europe, and that trend is expected to continue. The numbers can be mind boggling. A prominent private equity house announced, at the start of October 2024, a $10 billion investment in a single site in the UK. 

This kind of growth needs infrastructure. It's tempting to say that the key issue is identifying potential development sites with a high quality and reliable power supply. But the infrastructure demand is so high that identifying a development site with any suitable grid connection can itself be a challenge.  Data centres also undeniably have an environmental impact and ensuring this is properly managed will be necessary if we are to meet not only our data processing requirements but also our ESG targets. Whilst the recent designation by the UK government of data centres as ‘critical national infrastructure’ will mean increased government support in terms of anticipating and managing potential threats, if we are to meet both ESG targets and data centre growth requirements, a more pressing need is finding the necessary finance and ensuring effective recycling of development capital. 

It is here that securitisation has the potential to shine. Whilst recent deals in the US and in the UK (the latter a nascent market for data centre securitisation) have been met with much anticipation this has not been without some debate. What exactly is a data centre securitisation? Is it a CMBS deal or is it an ABS deal? What are the arguments on each side and, well, who actually cares? 

CMBS or ABS

Data centres themselves come in all shapes and sizes - well, all sizes anyway. At one end you have the Hyperscale centres (anything from 20MW upwards, often much larger), providing at least 10,000 square feet (930 square metres) of raised floor area and housing upwards of 5,000 servers. In the middle are the Enterprise centres, often custom-built to a tenant's specific requirements. And at the other end are the retail Colocation ('Colo') centres, smaller and multi-tenant by nature. There's also the growing fourth category of Edge data centres, responding to the growth of the ‘Internet of Things’ as well to potential use in the self-driving vehicle industry, where low latency is a clear must.

So a data centre securitisation is a CMBS, right? Not so simple. You have multi-tenant properties (a bit like big shopping centres, perhaps with a good anchor tenant too) and stand-alone bigger commercial properties, like a single-tenant office or large logistics warehouse (think SASB). This all feels quite familiar. 

But is it? Data centres are not just big warehouses waiting for servers, switching and storage systems to fill them. They require complex security systems (both externally and internally), significant uninterruptible power, connectivity and cooling, ongoing management and more. They do not need high footfall, voluminous parking or scenic views.  

If a tenant were to exit, for whatever reason, the (possibly only) sensible re-let is a replacement data centre operator. Depending on the type of property, this could be challenging (in theory, although current vacancy rates suggest that the challenge may not be so great). There are only so many tenants that require a 10,000 square foot data centre. If that centre could be converted into a retail colocation site perhaps it would be easier to fill quickly, and you can imagine a possible semi-alternative use arising. Converting it into, say, an office building or retail space would however require significant redevelopment and in a lot of cases may render the exercise unviable. Location may mean it is entirely unsuitable for anything else but a logistics hub. Shopping centres, hotels and offices, for example, all have very different location and design requirements to those of data centres - windows being quite high on the list of priorities for the former. Alternative use value has to be considered differently in the context of data centres than for other commercial properties. 

Alternative use is not the only way in which data centres may differ from traditional commercial property. The type of tenant also varies significantly across the types of data centres. In the AI sector in particular, newer, unrated, start-up style tenants may be more prevalent, and this may lead to a difference in the ratings that are obtainable. 

Revenue streams are often another key difference. For the smaller centres in particular, rather than lease agreements, clients may sign shorter-term service agreements. Whilst from a cash perspective they are all a revenue stream, the differences to the underlying contract require careful consideration. A tenant under a lease will enjoy different protections (and bear different obligations) than a customer under a service agreement. Leases often are binding against a lender and are site-specific. Service agreements, on the other hand, can be global in nature and transferable to different sites. 

It would be hard to argue that if you sell a bond backed by data centre revenues (and a handy commercial mortgage to boot) that you are selling commercial real estate risk. 

We asked “who cares” what type of deal is being offered to the markets. The investor, that’s who. Whether you are selling ABS, CMBS or widgets, if you don't know who your customer is, you're unlikely to be successful. You need to know your investor base. More than that, you need to know that what you are structuring is something they want to buy. Weighted average lives (WALs) differ between the types of securities - a traditional CMBS will have a WAL of about 8 years and esoteric ABS (which is where a data centre would sit) a shorter tenor of around 4-5 years (albeit rated to a far longer maturity). This diversification can be hugely beneficial to certain types of investors. Diversification of WAL and type of asset backing the securities becomes increasingly interesting. 

Many buy-side market participants are still learning the sector. They already know real estate. This is something different. 

Are we ready for lift off?

While data centres share many similarities with commercial real estate there are also many similarities with assets found elsewhere in the ABS markets. So, how should a data centre ABS be rated?  Current discussions often ask if  a data centre securitisation is an ABS (subject to ABS rating criteria) or a CMBS (subject to CMBS rating criteria), but we may be focusing on the wrong question here. There is variety within this growing area and the different types of data centre have differing inherent risks and different investor bases, and will not be the same investors who are looking for real estate risk. Are they investing in infrastructure? Undoubtedly? Will infrastructure investors be interested? Most likely. These differences will only continue to become more diverse as new technologies evolve. 

How, for example, do we classify a data centre that is orbiting the earth? That wouldn't exactly be real estate, but the combination of unfiltered solar energy and natural cooling surely make this prospect compelling (albeit one with different operational risks). It may seem fanciful to imagine orbiting data centres (but maybe not - think of sub-sea centres) but the inventors of new technology are rarely limited by a lack of imagination. Securitisation never has been limited by a lack of imagination either and perhaps that's why they can form such a good partnership. 

We embrace new technology every day. Sometimes it just doesn't fit into an existing category.  Just as different speed limits apply for different environments, pushing the market to accept data centre securitisation as either CMBS or ABS is surely going the way of the dial-up modem. 

If we keep at the forefront of our minds the end-goal, what an investor is going to want to buy, then some of these other points become much easier to manage. Whether CMBS, ABS or any other form of securitisation, the one thing they have in common is this: without an investor, you have no securitisation. If we start at the end, with the investor, questions like lease vs. service agreement, credit profile of tenant and even location all become much easier to answer. You want your team to be flexible, open-minded and inquisitive and to always have an eye on what the end-investor ultimately wants to buy. If we rush to categorisation, we may lose sight of what we are trying to achieve and so ultimately we should all care. 

Please contact Christian Lambie or David Shearer to discuss.