Blog Post

Rising Cost of Facebook Infrastructure; CTO Resigns

Stay on Top of Enterprise Technology Trends

Get updates impacting your industry from our GigaOm Research Community
Join the Community!

Last month, I wrote about Facebook’s insatiable hunger for hardware. Over the weekend, Spencer Ante of Business Week reported that Palo Alto-based social networking company had raised about $100 million from Triplepoint Capital, a venture lending operation. “It will be used entirely for servers,” Facebook Chief Financial Officer Gideon Yu told Business Week’s Ante. That gives us a sense of how much hardware is gulping down.

In other words, Facebook could easily increase its number of servers from current estimates of about 10,000 servers to over 50,000 servers, in order to keep up with its growth. Various web analytic companies estimate that Facebook has over 100 million visitors and serves up billions of pages every month. One of my sources had told me that Facebook was looking to spend about $9 million this year on hardware, but apparently, their needs are much bigger. All that spending should prove to be good news for Rackable (RACK), which has been a big server supplier to Facebook.

Facebook has built an enviable infrastructure. I personally believe that its desire to market Facebook Connect is a way to offload (at a later stage) some of the pressures Facebook apps are placing on its system . Regardless, there’s one exec who is not going to be around to help architect the next phase of Facebook’s back end bulk-up: CTO Adam D’Angelo has left the company, according to Kara Swisher and Eric Eldon.

Facebook spokeswoman confirmed the news and added that “D’Angelo will be transitioning out of his role as CTO and leaving Facebook to take some time off.” The company is currently looking for a VP of Engineering. Apparently, d’Angelo’s main interest was engineering, and the CTO role moved him too far away from that.

16 Responses to “Rising Cost of Facebook Infrastructure; CTO Resigns”

  1. If they’re going to be spending $100m every couple o’ years (as has been reported elsewhere, this debt is expected to cover server acquisitions for the next “year or two”) on just servers, and continue to grow, there’s not going to be much room for profit. This sort of thing is probably whey the IPO’s being avoided.

  2. @ Daniel,

    I think you are right about this – and that explains why Google spends so much money on its infrastructure.

    You should come to our conference though. It be fun to have you around.

  3. Daniel Golding

    A 23-year old, no matter how smart, is simply unqualified to run engineering at a firm like Facebook. Even if he has the development chops, he certainly doesn’t have the infrastructure experience needed.

    Of course, I’m waiting for one of your commenters to drop some piece of science like “why don’t they just use EC2” or “but what about the cloud?”. At some point, running your own servers makes sense, even if owning your own datacenter doesn’t. Thats the point Facebook is at.

  4. Pete Steege

    The virtual world is starting to look very physical. The Web kicks things off, but data “factories” are going to keep it going.