Monday, January 10, 2011

colocation long term fundamentals

Ken Baudry started an interesting serie of articles about the long term potential of the colocation sector at

Here are a few quotes from the first two articles:

>>There isn’t a single “data center market”.
There are multiple submarkets. Each submarket must be analyzed separately. An illustration is probably in order; Suppose that I told you that the average price of a new car was $30,000 and that I could buy a new car for $45,000. Is my $45,000 deal a good deal. Well, if I’m buying a $70,000 Mercedes then yes it is. But if I’m buying a $16,000 Hundai than I’ve kind of screwed myself looking at averages.


What all the analyst, investors, brokerage industry have tried to do is to say is that “its all one big market”. This is a big mistake that leads to poor investment decisions.


There are colo providers that sell plain old rack space, QTS. Providers that sell cross-connects; Telx and Equinix, Providers that sell managed service, etc. Don’t expect to buy whole space in Telx or Equinix, or in a managed services facility or in a retail colo facility at a competitive price, its not what they sell. Different target markets. Yes both WalMart and Saks are both retailers. No they do go after the same customers. Can you comparre them? Yes? Do they respond the same to changing economic conditions. NO!


>>Is there a danger of technical changes making today’s collocation facilities obsolete? Yes! But the bigger probablity is that changing technology will make office buildings obsolete! But let us put this in prospective.


Companies that exist on the internet, those that must be collocated with a major peering point or immediately adjacent to it are probably running close to the 5MW/Cab level, will continue to do so and perhaps exceed these densities.


Today only 20%+/- of all data center needs are being met by colo facilities. The rest being are located in single tenant owned or leased facilities. The power density in these centers average around 2kW/Cab (80 W/sf). Which brings me to my next point.

The colo industry has been very successful at attracting the “early adopters” and companies who don’t have the capital to invest. I believe that the lack of activity from enterprise customers, those not providing internet services, is largely due to fears about the economy but also because the colo industry has built a product that doesn’t fit the remaining 80% of the market. So in that sense, these facilities are already obsolete.

The threat to colo isn’t technolgy change (virtualization) or obsolesence . It is the failure to understand that this isn’t a one size fits all industry. To be wildly successful, the industry will need to do a better job at attracting the 80% that are currently on the sidelines.


To conclude, changing technologies will not make facilities obsolete, if management understands the concepts of submarkets and how to reposition the assets. On the other hand, changing technologies might make some management and colo operators obsolete.

No comments: