Cringley on Apple: The Math Needs Some Work

There’s an old saying about computers: garbage in, garbage out. The saying applies to an analysis of the new Apple data center by tech columnist Robert X. Cringley, who asserts that the facility in Maiden, North Carolina is big enough to contain 7.2 million servers, but is basically empty.

“I drove over, took some pictures, and talked to folks at the convenience store down the road,” Cringley writes. “My conclusions from this unscientific research is that the giant Apple facility is mainly empty. It’s a huge building filled more or less with nothing and why Apple built it that way frankly escapes me … I think it’s a joke. The building is a near-empty facility built primarily to intimidate Apple competitors. And so far it seems to be working.”

Wow. Did a well-known tech columnist just allege that Apple built an enormous fake data center? And, in essence, accuse Steve Jobs of modeling data center vaporware when he showed off pictures of server rooms filled with gear at the recent WWDC?

There are fair questions about why Apple has built at such enormous scale. But Cringley succumbs to a temptation that often strikes tech journalists: the urge to extrapolate a single data point into a broad conclusion about a data center. This usually involves either how many servers a facility can hold, or how much energy it uses. This process rarely turns out well, and Cringley’s effort is no exception.

Cringley’s estimate that the Apple facility could hold 7.2 million servers is based on the notion that Apple has 1 million square feet of space that can house racks of servers. The problem: The Apple data center is 500,000 square feet, not 1 million (a fact that has been widely reported). The equipment area – the only space that can actually hold racks – makes up just 184,000 square feet of that space, according to planning documents (See our Apple Data Center FAQ for more details).

Cringley also underestimates the amount of square feet required per rack, as noted by commenters from the industry. As a result, he wildly overshoots on his estimates of the number of servers and storage units Apple could house in the building. His conclusion that the building is empty is apparently based on seeing only one truck enter the facility over the course of an hour.

To be fair, Cringlely has a history of much better work on the data center sector, particularly his early scoop on Google’s use of shipping containers to house servers.

Has Apple overbuilt? That’s a better question. It has committed to spend $1 billion in North Carolina, but has nine years to do it – meaning the real answers to the questions about Apple’s data center will be answered over time.

Get Daily Email News from DCK!
Subscribe now and get our special report, "The World's Most Unique Data Centers."

Enter your email to receive messages about offerings by Penton, its brands, affiliates and/or third-party partners, consistent with Penton's Privacy Policy.

About the Author

Rich Miller is the founder and editor at large of Data Center Knowledge, and has been reporting on the data center sector since 2000. He has tracked the growing impact of high-density computing on the power and cooling of data centers, and the resulting push for improved energy efficiency in these facilities.

Add Your Comments

  • (will not be published)


  1. Oroboros

    Seems like the journalist is trolling for traffic. Like a shock jock host getting ratings for outrageous remarks, Cringley is writing stuff to bait reactions like yours just to get hits.

  2. amadeo puzzo

    "Cringley succumbs to a temptation that often strikes tech journalists: the urge to extrapolate a single data point into a broad conclusion about a data center." I haven't ever heard or read of a tech journalist succumbing to the urge to extrapolate a single data point...about a data center. Is this urge really that specific? Is it endemic to tech journalists? Or might I, for instance, succumb to extrapolations about data centers? Gee, I hope I don't.

  3. Amadeo: Succumb can be defined as "failing to resist" as well as to "die from the effect of a disease or injury." I've seen writers from The Wall Street Journal and New York Times seek to use facts we've published at DCK to make calculations about entire networks of data centers. I've had many other journalists call to ask whether there was a formula they could use to say how many servers were in a data center of a certain size. Many simply ignore admonitions that data centers are different and the math doesn't work that way. Their editor wants to see a number, so they come up with one. Companies like Apple, Google, Microsoft and Facebook don't talk about how many servers they have, so it's a freebie - journalists can come up with any number they like, and the company will never confirm or deny.

  4. So if they have 184,000 square feet of space, and they use racks that are 22-inch-wide by 30-inch-deep, and each rack has a 30 inch hallway in front of it, and they reserve 20% of the building space for walkways, and they build their racks 15 feet tall (such as in Bell COs), and they use 9 rack-unit server chassis (such as the IBM H BladeCenter chassis), and you need an extra 2 rack-units of space for each chassis (e.g., for cable management, power supplies, etc.) and each chassis hosts 14 servers, then they should be able to support 2,023,308 blade servers. If instead they use 1U servers, but they're half the depth, and you reserve 10 Rack Units per rack for network gear, then you can fit 92 servers per rack, and you gave 32116 racks, i.e., 2,954,672 1U servers.

  5. Mark: Interesting math. I think the unlikely variable would be the 15-foot-tall racks. Data centers typically use a seven-foot rack, and while we've recently started seeing racks as tall as 9 feet, a 15-foot rack seems unlikely. The racks in the data center photos Apple displayed at the WWDC event are definitely not nearly that tall.

  6. Shane

    “Cringley succumbs to a temptation that often strikes tech journalists: the urge to extrapolate a single data point into a broad conclusion about a data center.” Forgive me, but I believe you just used a single data point to extrapolate a broad conclusion about Cringely.

  7. amadeo puzzo

    @Rich, I think you're missing my point. The sentence starts out well, but to say that all tech writers end up extrapolating about one specific thing as esoteric as this is silly. Get your head out of your...erm...datacenter, and look at your grammar.

  8. Well this article indicates it may be double the 500K sq ft that was 'planned' ... but yeah 7FT/40U racks that are lets be generous - 30" wide x 48" deep + space in front & back for cooling, and people moving around ... and what 1U MP (lets be conservative and say 8 core) severs running over maybe 10Gbps fiber/network ... but now we need space for the cabling, the RAIDs/NAS/SANs ... and the power and the cooling and the place for the two administrators to work, eat and sleep. I'm thinking 1 billion buys Apple a whole helluva lot. Maybe Apple will regain their sanity and bring back OS X Server and the XServe? Maybe they will bring back matte display monitors? I wont hold my breath, but Thunderbolt looks kinda cool. O yeah, I bet this would cost the Federalies about a trillion USD$, not that the USD is worth much these days. So who's gonna default first, the Euro or the Greenback?

  9. Stiffy O'Riorda

    The thing you're all missing is the space for the giant green lizards. And you call this responsible journalism? Jeez.

  10. Eric

    What's remarkable here is that such disbelief could be expresses about Cringley's reporting when he has done such link-baiting nonsense for years.

  11. Bob

    Having been in there, he is way wrong about being empty. The facts that I have seen on this website for the facility are so "guess work" it is rediculous, Greenpeace was way off and everything I have read is all based of of incorrect numbers or based off of the proposed two buildings, not the one that is present.

  12. Adam

    I don't know why people are complaining. I found the article entertaining, and enjoy the idea that some reporters are in fact dangerous, mad-as-a-march-hair-crazy who see conspiracy theories where other people seen otherwise mundane lack of information. :)