In the last two days, the New York Times [subscription required] and Wall Street Journal [subscription required] have run stories on how Microsoft, Google, and Yahoo! are competing for cheap electrical power to deliver services such as web search, e-commerce, and web video hosting. It seems ironic that a key factor limiting growth in the virtual world could be something so physical.
Could this battle knock out any of these competitors? I don't know the answer but investors should monitor its progress. Unfortunately, these companies' financial statements don't yield insight into the capital and operating costs for the data centers. Google's most recent 10Q notes an important risk factor -- some of its data centers are located on earthquake faults!
But investors should not be concerned that data center costs are out of control. Each of the three companies generates free cash flow -- which represents cash generated by the business after capital expenditures. However, between the end of 2005 and the first quarter of 2006, Microsoft's and Yahoo's increased but Google's tumbled:
-
Microsoft + 78% to $7.3B
-
Google - 81% to $292 million
-
Yahoo! up from -$396 million to $249 million
The factor limiting growth is not money but the ability to find new locations with access to cheap power. To deliver these services globally on a 24/7 basis, Microsoft, Google, and Yahoo! operate massive server farms. These football field sized buildings filled with thousands of servers networked together have massive appetites for power -- the WSJ notes, "one large data center can consume enough juice to power a small city of 30,000 to 40,000 people." The WSJ goes on to point out that cutting as little as a penny per kilowatt-hour from electricity rates could reduce annual expenses by "millions of dollars."
If I was inclined to invest in raw land, I would certainly be trying to anticipate where in the world these companies might be looking to build huge facilities located next to cheap sources of power. As Google did in The Dalles, Ore., it would make sense to look for abandoned factories that had previously consumed power near a dam.
And I would not just consider the US. It may well be that there are enormous sources of cheap hydroelectric power in countries like Africa, China, and India. If these companies could find a way to build, operate, and transmit globally the information in data farms in these countries, investors could make some impressive returns by anticipating such location decisions and buying the land ahead of time.
Of course, there is another investment opportunity in technology which would lower dramatically servers' power consumption per unit of information processing power. Any server manufacturer that could introduce new products which required less electrical power would see a surge in orders.
I don't know where in the world the cheapest electrical power is available for locating huge server farms. But I have little doubt that these companies' demand for more processing power will grow as services become more processing and bandwidth intensive.
Investors will profit if they can anticipate and invest in server farm locations that would appeal to Microsoft, Google, and Yahoo!. Investors will also score if they invest in companies that provide technologies that yield servers with lower power consumption per unit of information processing power.
Any suggestions?











Reader Comments (Page 1 of 1)
6-15-2006 @ 4:23PM
ron seredian said...
Data center power is just one aspect since cooling it draws an enormous amount of power too. I think it's time to give DC power another look. DC eliminates the double or triple conversion that the present system demands. Flywheels are another area that hasn't been looked at seriously. These power generators can and will be an important alternative to the present method of AC power being converted to DC (for the UPS or back-up system) then back to AC. Not only inefficient but a lot of heat is given off with the present way of doing things so let's not move to India or China just yet, let's make the whole equation smarter.