For web users concerned with their privacy, Facebook’s decision to file for its initial public offering late last month probably wasn’t good news. When the S-1 filing hit, I wrote a post for GigaOM explaining the connection between Facebook’s large and growing infrastructure budget and advertising, the latter of which is the primary culprit behind many privacy concerns. The gist is that ad-supported web platforms such as Facebook have to spend lots of money on infrastructure in order to maintain adequate performance levels and power new features. However, that means every dollar spent on infrastructure adds up to even more money brought in by advertising. Investors, after all, like income.
Facebook made clear just how big a part of its business infrastructure is in its S-1 filing: $33 million in 2009, $293 million in 2010 and $606 million in 2011. It plans to spend between $1.6 billion and $1.8 billion in 2012. Most likely, the 2011 spike was related to the construction of its first data center in Prineville, Ore., and possibly the acquisition of space for new ones in North Carolina and Sweden. The spike in 2012 probably will come from building those facilities and filling them with gear. But when compared to other companies with large web operations, Facebook’s spending doesn’t look so high.
As the chart above illustrates, Apple,* Google** and Microsoft*** all invest much more in capital expenditures — and generate much more money — than Facebook does. However, there is a big difference in how these companies spend on infrastructure. Whereas Microsoft and Apple spent an average of 3.3 percent of total revenue on capex between fiscal years 2009 and 2012 (thus the skinny lines), Facebook, Google and Yahoo**** spent an average of 10.13 percent of their revenue on capex during the same time frame (the fat lines). Presumably, that has everything to do with revenue streams. Apple and Microsoft make lots of money on hardware and software licenses, while Facebook, Google and Yahoo don’t have that luxury; they rely largely on advertising on their free web services. And as Facebook explained in its S-1:
Our business is dependent on our ability to maintain and scale our technical infrastructure, and any significant disruption in our service could damage our reputation, result in a potential loss of users and engagement, and adversely affect our financial results.
Part of this is standard S-1 risk-factor fare, but it is also a frank assessment of the state of Facebook’s situation. If it slips up, someone — be it Google or some innovative startup — will be there waiting to steal users’ attention. This puts the latter group of companies in an awkward position where they have, essentially, three choices when trying to make enough money to support their infrastructure habits. They can 1) get more advertising revenue, 2) get into new revenue-generating businesses or 3) start charging for their services.
Although it is not necessarily a bad idea, users might not go for No. 3. Getting into a new business — such as selling hardware or paid services — is no easy feat, although Google appears determined to do it. So No. 1 it is, although this almost certainly means even more analysis of user data, as advertisers expect better-targeted ads to justify choosing any given platform over the myriad options available for online advertising. Just look at the privacy shellacking Google has been taking lately over its decision to aggregate users’ data from across Google services. From a financial perspective, the chart below is illustrative of the conundrum. Capex as a percentage of income (pretax) is significantly higher for web-only companies than it is for those with actual products to sell. Building new infrastructure, as well as the costs of operating it, can take a major bite out of a company’s bottom line.
From a business perspective, Yahoo is perhaps the poster child of what can go wrong when ad revenue dries up. Yahoo is doing plenty of traffic and fancies itself a technology leader, so it has to keep spending on infrastructure. But the cost of obtaining new infrastructure (more than 70 percent of Yahoo’s actual profits — not to mention operating it — looks to be an insurmountable burden. It can’t start selling users’ personal data to advertisers, so Yahoo brings in more traffic and more data to analyze by producing news content that, of course, readers don’t pay for. But infrastructure spending and ad revenues are a cyclical model: If advertisers still choose to spend with Facebook, Google and other sites, Yahoo’s costs of scaling to meet traffic and data-management requirements might become too much for investors or the bottom line to bear. Without ad revenues, there can be no new infrastructure and vice versa.
Bringing the discussion back to Facebook, there are two primary paths the company can take to grow its business. One is to become more like Apple or Microsoft and offset its infrastructure costs by selling high-margin products or possibly even charging for its service. Being free is great to attract early users, but it is not a business model that always scales well. Contrast Yahoo’s plight with Google’s mission to charge money, exemplified by its growing, but still relatively small, paid-service business and its forays into hardware.
Facebook’s other option is to maintain the status quo, at least for a while. That means more spending on infrastructure to support more traffic, more data and more features. This will be paid for by new advertising programs and platform upgrades that promise to put the right eyes in front of the right ads, games and services. It might not make privacy hawks happy, but it is life in a world where users expect a free web that actually works.
* Derived from financial-statement line item “Payments for acquisition of property, plant and equipment”
** Derived from financial-statement line item “Purchases of property and equipment”
*** Derived from financial-statement line item “Additions to property and equipment”
**** Derived from financial-statement line item “Acquisition of property and equipment, net”