It’s been a year since I wrote and published the blog titled “An ‘open letter’ to data center investors” and I think it is safe to say that it caused quite a stir at the time having been viewed over 4000 times within a very short space of time.
In fact I found myself at conferences where over lunch people would say “Oh!…. you’re the guy that wrote that blog!”, I didn’t really need to ask which blog they were talking about, it was always the same one! Personally I think I’ve written much better blogs but that particular one hit a nerve albeit within a relatively small business community.
Causing debate
Pretty much everyone that I ever spoke to about the blog said they agreed with the ’truths’ I’d spoken about within it even if their company line was that they didn’t agree with many of the points I’d made. It certainly caused some debate, a number of blog responses and ultimately did what I wanted it to achieve and that was to start the conversation about how our industry was changing.
So my marketing folks recently reminded me it was coming up on a year since I wrote that blog and that I should write a follow up or ‘look back’ blog on the anniversary and see what if anything has changed. While I wasn’t exactly predicting the future with it, looking back at it and comparing some of the statements I made with where we are a year on sounded like a good plan, so here it is!
Let’s go through it point by point and see if anything has changed and if my own take has changed a year on.
1 Data Centers can’t be valued as a property business (any more)
The primary point I was trying to make here was that the real estate wasn’t the biggest cost on a long term TCO basis and that operating a data center business like a real estate business was not the way of the future.
In this past year there has been a huge amount of M&A activity in this market which is always a great way of crystallising the valuation of any business (or those around it). However if you look closely at the valuation metrics of deals past they vary greatly, perhaps more than one might expect for a commoditised business.
I could go into some examples but we’ve a lot to get through here - it was a long blog - so I will end this point by saying, go and do your own analysis (look at deal data from folks like Evercore) and draw your own conclusions as to why seemly like parties get very different valuations and then ask which aspects are easily replicable and which are not (or more fundamental to the way they run their business).
2 Data Centers are not “The Cloud”
Really I cannot believe I still have to explain to people that - yes - the data center is a foundational building block that enables “The Cloud” but that they are not on their own “The Cloud”.
Cloud in economic terms is the commoditisation of IT services, meaning that delivery at scale, low cost, low margin and often highly competitive. Whether any particular data center is a good (or bad) building block for the enablement of Cloud services (or not) is much more a function of its Total Cost of Ownership and Unit Delivery Cost than it is about whether it’s certified by the Uptime Institute (or anyone else) or not.
That’s not a bash against Uptime or any other certifying body, by all means get your data center certified, today it is almost a license to sell capacity into the colo market (not enterprise or cloud capacity though).
From an investment perspective, much more important a question is “is this asset of the right class based on the rental I can expect as more loads move into the Cloud and thus how competitive is this asset going to be both now and in the future? And what reinvestments might be needed to be made over the asset lifetime in order to keep it competitive as the market demands shift?”
3 The Cloud is not your friend it’s your competition!
This was a specific statement for the pure colo guys out there. It was a message that was meant to stimulate out of the box thinking about how they might partner with their Cloud customers rather than just see them as customers who always want the lowest price per rack as possible and say things like “I don’t need generators”, etc.
Thankfully I’ve seen many examples over the last year of colo’s that have partnered with their Cloud cousins and hopefully in a way that’s win-win for both parties. The most obvious way is a revenue share model where the colo gives the capacity for a lower cost (or maybe zero cost) but takes a cut of the Cloud company’s revenue instead.
Of course very few colo’s out there are likely to get the tier 1 Clouds to go for such a model but even for the small guys I believe this is a better longer term model to keep Cloud operators in your colo site and away from building their own (at a much lower cost) once they get bigger - and then they really do become your direct competition.
I think over the past year I’ve seen most (not all) colo’s get this right or at least make the right noises in the marketplace about their strategy on this point.
4 Focus on revenue and growth and it will all be fine
Well, I think we all know that double digit growth has gone in the tier 1 markets, competition is greater, despite what market capacity indicators might say, there is plenty of capacity both now and coming on-line or even sitting in the wings.
The players in colo land have consolidated and continue to. Colo’s being dedicated towards wholesale or retail as a business model has been greatly eroded (many do both now) and if you’ve not realised yet that you need to pay attention to the bottom line (as much as the top line) then you are doomed, maybe not immediately, but really your long term prospects are bleak!
Also people are always naturally focused on new projects, new capacity and new markets, but really spending time and effort to also look back at the economics of existing (and often older) assets in a much more granular way can yield results that allow you to improve the performance of those assets and also not make the same mistakes within the new ones too.
I’d add a message to those that now have ‘rich parents’, don’t for a moment think that just because you have rich parents you don’t need to put as much focus on costs as others that don’t. Being competitive in the market is predominately a function of your cost to deliver service to each customer. If you don’t know that or don’t have it under control or it’s out of whack with others in the same geographic market as you then I encourage you to get a handle on it really quickly.
5 Growth in data centers is a function of the explosive growth in technology, mobile communications, globalization, etc.
Colo is STILL a transitional step between legacy enterprise in-house and Cloud (hybrid or public). People will tell you there’s no need to panic now as although the enterprise is starting to move away from legacy applications that are tied to legacy hardware and legacy networks and it is going to take time for that to really bite, well, if I were you I’d panic now if you have no Cloud strategy!
I often argue with people who are the naysayers of the Cloud that always cite me the most critical business application as an example of why Tier 4 enterprise data centers will never die.
Really! I mean come on, I never argued that it was an either/or situation. My point has always been that while the macro level growth is there and long may it continue, the way the demand is serviced will continue to shift from traditional IT (physical) to Cloud based IT (virtual). As a colo only operator, if you don’t have a good story and solution for the CIO around their gradual move to Cloud enabled apps then again, you’d better get one together reasonably quickly.
Remember while Cloud is accelerating the commoditisation of IT services, so in theory, lowering the cost to deliver business applications for the corporate CIO, often, it is the Agility that’s gained by shifting their legacy applications to Cloud technology that’s the bigger driver. (I could write another whole Blog on this topic alone!)
6 My product and my data center business are unique and thus immune to commoditization
If you are a colo and STILL thinking/believing this then may I wish you good luck for the future.
I’ve seen a lot of great and successful data center businesses out there and they could all easily say that in some way they offer a unique aspect to their customers that differentiates them and is the reason why they are successful.
In general I agree, they do all have aspects of unique differentiation, but the reason I chose those three examples is that it’s easy to see they all look quite different and yet the fundamentals of their business and product are the same.
Thus all data center businesses need to pay close attention to being competitive (in each market) and having the right planning and tracking tools to delivering their core product in a much more flexible way and against a much more granular cost model and then use the unique aspects of their businesses to win against the competition.
7 Customers will stay despite the costs because moving is hard and costly
Like my statement in point 5 above, the enterprise market have their own technology shift to go through and that will still take some time for the majority. However more and more are shifting their development to Cloud friendly and Cloud ready application architectures meaning customer stickiness from a physical perspective will continue to slowly diminish over time.
With things like Open Compute and new technologies like Vapor in the market the next few years I believe we will continue to see an acceleration away from the traditional tie between hardware and software.
8 For most customers renting space in a data center is cheaper than owning a data center
Well the economics haven’t changed over the last year enough to make Colo (sub 1MW) cheaper from a TCO perspective than building your own in-house. However most enterprise businesses are not experts (often nowhere near!) at actually operating a data center, so it’s can become a royal pain in the behind and worse still the risk and increased cost of screwing it up could end up negating the lower TCO.
Again, this isn’t about an all in or all outsourced issue. It’s combination of what do the three major classes of applications in my business require from a cost, criticality and agility perspective and then matching those against all the options available to you.
Really, building it yourself (meaning, managing the build yourself) is a bit of a fools game these days, unless you are doing it at scale yourself (Google, Facebook, Microsoft, etc) then you’re better off paying what should be a relatively small premium to a firm that’s productised their offering and then making sure you aren’t their first customer.
Conclusion
Well I guess these have changed a little from a year ago.
For the colo’s out there getting a solid story and capability around Cloud makes sense, if you are an enterprise and you are still trying to build your own data centers, well, I’d say, think really hard before you do any more but if you have someone else build it for you then ask to see at least the first and second generation product in the field to see what they learned and fixed as they went forward.
For the data center investors - for which this is primarily targeted - some of you are invested in really good businesses today. I’d make sure that the folks you have running those businesses have realised the shifts taking place in the market and are planning and reacting accordingly. Nobody likes surprises!
Finally, and really I shouldn’t need to say this but I do given what I see out there everyday! If your operating company can’t tell you what their delivery cost per customer is then they need to get to that pretty quickly, because not knowing in today’s market leaves too much margin for error.