Blog post

Who is leading in cloud? Data and context matter…

By Douglas Toombs | November 04, 2013 | 2 Comments


Earlier today, IBM launched a “bold” (their words) new marketing campaign about cloud computing, with an ad in today’s Wall Street Journal stating that IBM’s cloud “powers 270,000 more websites”  than Amazon.  Not too shortly after the ad was in the market, I received an inquiry asking if this figure is actually true.

The short answer to that question, is yes.

The medium-length answer is that IBM is quoting figures from which appears to crawl through the top several-hundred-thousand websites on the Internet and figure out where they are homed.  The methodology probably uses would involve verifying IP addresses for the top sites against various registries like ARIN, RIPE, APNIC, etc.  I’ve actually performed similar analysis of the market in the past so I believe the methodology (and the results) are generally sound – for what is being stated.

The long answer (because the devil is always in the details), is that the types of sites matters.  About 18 months before their acquisition, Softlayer was (according to public statements) on a revenue run-rate of about $340 million annually.  However, in the same quarter for Amazon, their “Other” revenue was coming in at a run rate of about $1.6bn annually.  While the “other” line item in Amazon’s financials also includes some non-AWS items – like marketing and promotional agreements, seller sites, and co-branded credit card agreements – those portions of “other” are not generally believed to be significant, and Gartner believes (as many financial analysts have published) that Amazon likely had close to $2 billion in AWS revenues in 2012.

So how does IBM/SoftLayer end up being the provider to so many more “sites” on the Internet, on presumably smaller revenue?  In the world of hosting, size matters – and SoftLayer had good successes in the market in underpinning some shared hosting providers, such as HostGator (which claims to have 12,000 servers under management, and is a partner of SoftLayer’s).  This can tend to skew numbers a bit, as shared sites are frequently quite small – in fact, even as I am writing this blog post there is an ad for “$0.01 Web Hosting” with HostGator running on’s site.  Shared hosting is not “cloud”, and what small websites lack in size … they tend to make up for in volume as the “long tail” of the market.  To be sure, using the same yardstick that IBM has chosen, GoDaddy nearly surpasses Amazon and SoftLayer combined with 431,087 websites as of today … compared to 358,532 for SoftLayer and 73,821 for Amazon.

Based on Gartner’s own research (using similar methodologies as we believe that Amazon Web Services is now being leveraged by 10% of the top 5,000 websites in North America – so Amazon is definitely taking a bite out of the market from many other players. 

Of course, none of this diminishes the value of the property that IBM has picked up – SoftLayer has some very innovative capabilities, all built with in-house developed intellectual property, and was very successful overall in the broader hosting market … and will likely continue to be in their new home with IBM (my colleague Lydia gave an in-depth overview of the transaction here). However … that doesn’t make for a very eye-catching headline to splash across the pages of the WSJ – so while the marketing team over at IBM is being accurate; they’re choosing a very specific view of the market as well.

The Gartner Blog Network provides an opportunity for Gartner analysts to test ideas and move research forward. Because the content posted by Gartner analysts on this site does not undergo our standard editorial review, all comments or opinions expressed hereunder are those of the individual contributors and do not represent the views of Gartner, Inc. or its management.

Comments are closed


  • Raj Dutt says:


    I’m glad that you acknowledge that the marketing team at IBM is being accurate.

    At the end of the day, I totally believe that Softlayer “powers more websites” than AWS.

    To me your article boils down like this:

    Your article says:

    Short Answer: Yes
    Medium Answer: Yes, by a reasonable methodology
    Long Answer: Yes, but they have a lot more tiny sites.

    This passes my internal marketing-bullshit snifftest (believe me, not necessarily easy to do) with ease.

    So why serve up a side of “but its not really cloud” with your otherwise congratulatory note?

    Shared hosting is not cloud. Sure, but Softlayer doesn’t do shared hosting. Neither does Amazon. They both sell raw compute (servers). The little long tail shared hosting sites are not Softlayers customers.

    If Amazons customers had included big shared hosting companies from the get go maybe you’d be writing what a “cloud victory” this was instead of saying “shared hosting is not cloud”

    People forget that when AWS first launched it was a flat out nonstarter for hosting websites; its sweetspot was for backend and batch processing. Even today it’s a bit wonky (from a fundamental design/constraints/cost standpoint) to do a lot of shared hosting on.

    The reason that companies like HostGator use SoftLayer instead of AWS is because SL offers a vastly superior value (lower price) and experience (capabilities) for their workload.

    The shared hosting (of a lot of small sites; long tail) is what is going to boost Softlayers stats. They had a head start, and they have a more appropriate IaaS platform. So they ended up powering more sites.

    Totally fair.

    On the flip side, I wouldn’t be surprised if the top 10 AWS customers combined approaches or eclipses Softlayers entire revenue.

    Another argument I’m hearing is that since most SL servers are bought on a monthly basis (and are bare metal) then it’s somehow ‘dedicated servers’ instead of ‘cloud’. I think this is a weak argument since their platform is so highly automated. They even offer their customers the ability to spin up additional servers on-demand, on an hourly basis.

    At what point do they not become a cloud? If Amazon offered servers on a monthly basis would that not be cloud? They already offer Reserved Instances – doesn’t violate the “no capex pure opex” rule of the strict definition?

    There’s been a general undertone that somehow Softlayer ain’t cloud, and AWS is. I don’t really like it or agree with it.

    It sounds like I work for SL but clearly I don’t. I just respect those guys a lot; we competed back at Voxel and continue to do so at Internap.

    I could also be writing these exact words about Internap which makes it an easy way to do a plug and hopefully add to the discussion 🙂

    Anyway, my $0.02, keep up the good work,


    Raj Dutt
    founder Voxel dot Net
    SVP Technology Internap

  • Lydia Leong says:

    Raj, it matters in terms of what capabilities are actually available to customers on a cloud basis. SoftLayer’s dedicated hosting options are much broader than what’s available in their CloudLayer. That in turn affects the use cases, and it’s exacerbated by the fact that the SoftLayer’s capacity pool is MUCH smaller than AWS. I can think of clients who have had use cases where SoftLayer’s bare-metal approach would have been preferable from a performance standpoint, but SoftLayer couldn’t support the scale and capacity flexibility they needed, while AWS could do it trivially.