By David Gross
The downs and ups we've seen over the last month in data center stocks have given a good indication of how investors can be forced into manic buying and panic selling in this industry. But in order to avoid mediocre returns from buying and selling with the herd, investors need to stop doing the following:
5. Worrying About Shrinking Servers Reducing Demand
Wall Street seems hung up on this, while few in the industry are. And for good reason. One is that data centers are filled with switches and routers that are not getting any smaller. Many common boxes, from the Cisco 6509 to the 7000-series routers to F5's load balancers, are the same size now as they were five years ago. And while servers are unquestionably getting smaller, they are not doing so at a pace that will destory demand for colocation. Rackspace, for example, has .36 servers per square foot now. Two years ago, it had .29. Additionally, there is far more data center space being leased now than there was ten years ago when commercial blade servers didn't exist, and corporate data centers were filled with monstrous Sun E10000s.
4. Looking at Supply/Demand Conditions too Broadly
Speaking with institutional investors, I get asked a lot about general supply/demand trends. The problem is that this is the same question everyone else asks, and does not lead to any conclusion that will give you an edge. Moreover, there is no general market for data center space. There are certain industry, price, and cross-connect requirements in Northern New Jersey, which differ dramatically from the buying industries, price, and cross-connect issues in Dallas, which differ dramatically from conditions in Los Angeles. But none of this gets considered with very high-level ideas about "where demand is going".
3. Fearing that Colocation is About to Become a Commodity
Many investors are afraid that colocation is about to become a commodity. But they have nothing to worry about, because it already is a commodity. As are semiconductors, CDNs, Ethernet Exchanges, and just about any other telecom service or computer hardware product. Whether or not colo is a commodity is irrelevant, what matters is whether a certain provider has high market share and therefore can maintain margins, just as Intel has done in chips and Akamai in CDNs, even though prices in both cases have been falling through the floor.
Market share in colo and wholesale data center space is only relevant on a region-by-region basis, making it even more important to stop looking at supply/demand at too broad a level. Equinix, for example, is remarkably strong in the DC area, but has much lower share in Los Angeles. Pricing and desirability as a cross-connect location will reflect this. Moreover, an extremely well-capitalized startup with half a billion dollars is only going to be able to focus on one or two markets at best, reducing the ability of some garage company to come in and take significant share away from an established supplier.
2. Overreacting to Changes in Revenue Guidance
I track an expectations multiple of guidance changes to stock price swings, and it has been simply ridiculous since Equinix's October 5th warning. On October 6th, the company's stock lost 35% of its value off a 2% guidance drop, while on October 27th, it popped 7% on a .4% increase in guidance, for an expectations multiple of 17. This week, Terremark jumped up 13% on Tuesday after lifting guidance 1% on Monday. Similar swings have occurred with data center technology suppliers such as Mellanox, Riverbed, and F5.
I'm sure some traders have figured out how to make money from the manic buying and panic selling, but fundamental investors who respond so sharply are setting themselves up for weak returns by getting in and out at exactly the wrong times.
1. Overemphasizing the Importance of the Macroeconomy
This industry was in a depression in 2002, when the rest of the economy was in a mild recession, and it grew in 2008 and 2009, when the rest of the economy was in a deep recession. Data center revenue growth has shown a remarkably weak link to GDP growth, because it's tied to growth in web usage, Internet traffic, and financial market data and trading volumes, which have all grown when real GDP hasn't.