Hedge fund manager Jim Chanos has made headlines again. This time, he is betting against ‘legacy’ data centres with a thesis that is underpinned by a dynamic that I don’t recognise in our industry.
Well, ‘IMHO’ some of the assumptions that have informed his ‘big-short’ position lack the nuance that comes with sector-specific knowledge. Value is accruing to cloud companies, as he correctly suggests, but that does not mean that it is diminishing among bricks and mortar that underpin this.
Chanos proffers that the data centre industry is at odds with its key customers – which I can only assume are the likes of Google, Amazon and Microsoft – and that these ‘competitors’ are set to turn on providers and eat their digital lunch.
He points to facilities that have been built by the hyperscalers for their own use. However, there is a fundamental question he has failed to address: does this construction wave actually indicate a trend, or is it far more nuanced than that? In the latter case, these assumptions need to be debunked.
All data centres are equal and some are more equal than others
The global data centre sector is not uniform and neither are the requirements that must be fulfilled. The same applies to portfolios within REITs. There are ageing facilities within wider portfolios that are going to need significant CAPEX investment to bring them up to scratch, but these facilities are not indicative of the sector and are often in key locations that will justify further investment.
Microsoft can afford to invest in land in Sweden for an efficient cooling system that hits sustainability targets, but such a site is not practical for everyone. The roles of all data centres are of course changing and adapting to cutting-edge technologies that come onto the market.
Companies such as Kao Data, for example, are looking into how designs and structures can best support future technologies: will further automation make the service more efficient? Can centres be dark? These are the questions that will shape their future.
The concept may evolve, but the practice is not outdated.
Keep the main thing, the main thing
It is pretty outlandish to assume that owning and fully operating the assets in every territory is important to being a provider of global retail cloud capacity. While there could be exceptional cases, such an overarching assumption is unlikely to be a good use of capital for businesses. Data centres are providing a service and they are completely on top of that service. We are consistently investing to ensure that the tenant model continues to bring that value add that is the hallmark of the sector.
As we all work towards meeting sustainability goals, we are investing and ensuring that the service we provide is of the best standard; keeping the main ‘thing’ that we do well, the main ‘thing’. A company like Microsoft is obviously capable of operating at the highest levels and digesting the costs that come with doing so, but such levels are unattainable for the vast majority of companies, not their expertise and not their priority. Data centres are a whole service; not just rental space.
Hyperscale today doesn’t mean hyperscale tomorrow
Google and Co are scaling up, undoubtedly, but such a trajectory comes with no guarantees of consistency in the future. We have use cases in healthcare, automotive, the Internet of Things and other sectors that are ramping up in demand: cloud suitable or cloud sensible data sets which are coming from businesses with no desire to own their facilities. The ‘Big Three’ may give up their tenancies in favour of buying their own places, but for many another renting remains the best option.
High performance compute = high returns
When anything becomes commoditised, it has an impact on value. However, the complexity inherent in growth sectors such as artificial intelligence, virtual reality and high performance computing protects value and creates more headroom for growth.
The marketplace is developing rapidly, partly as a result of the pandemic, sure, but also due to large demand for high performance products. Demand for data centres is not slowing down.
Through a legacy US-centric lens, there may be a rationale to Chanos taking such a hedge. Yet on reviewing his arguments and thesis, I cannot share his opinion.
’IMHO’, there is a strong future for data centres and exciting opportunities to come in this space as our world continues to become more digitalised. While the hyperscalers will always build, it is equally true that they will be partnering with operators around the world for many years to come.
In many ways, perhaps this will create some hiatus in the investment market that will allow Goldacre to do what it does best – invest in the teams that are building the future.