Rich Miller says 2006 was a huge year for data center stocks. Check out his cool chart:

In contrast, let's take a look at how mass marketing hosting stocks did over the past year (out of laziness, I'm using 12 Jan 06 - 11 Jan 07 stats):

* Web.com: UP $0.14 or +3.33%
* Tucows: DOWN $0.05, or -5.38%
* WebSitePros: DOWN $0.90, or -8.91%
* United Online (parent company of FreeServers.com, BizHosting.com, GlobalServers, and MySite.com): DOWN $1.01, or -6.69%

One possible explanation for this discrepancy is, even as SaaS and social networking providers fill up Savvis' and Level 3's data centers, they're reducing the need for end users to maintain web hosting accounts.

I know that Lou from HostMySite will disagree. A few days ago, he commented on my other post that we're too caught up in our limited view of the Internet world. While the likes of MySpace and Flickr have impressive numbers, Lou's not sure they're widely accessible to the common person. He sees "a vast opportunity that lies outside the existing community of advanced web users".

In which case, it seems our battle is against not SaaS and social networking, but Google and Microsoft.

(This chart was made with Zoho Sheet, using stats from Netcraft's latest web server survey. Microsoft and Google accounted for 46% of last month's hostnames growth.)

I learned a few things about Mirror Image today. Founded in 1999, it's one of the few dotcom era CDNs to have survived the bubble. 2006 has been the company's best year ever: 45% growth, contracts with major government agencies such as the Department of Defense and Center for Disease Control, great relationships with happy customers including Amazon, the New York Times, the Home Shopping Network.

VP Sales & Marketing Jim Hart (an old timer who's been with the company since 2001) tells me that Mirror Image maintains Content Access Points (CAP) in 23 major Internet exchange points around the globe. Jim says the CAP model is more efficient than Akamai's Edge Platform because with smaller pools of servers deployed across 71 countries, Akamai has less capacity for maintaining local copies of customer content. As a result, Mirror Image delivers a 98% cache-hit ratio (98% of end user requests are served from cache, rather than retrieved from content owners' home servers), versus Akamai's ~75%.

In addition to Akamai, Jim sees Limelight and VitalStream as major competitors. He hasn't run into CacheFly or BitGravity at all, and while he isn't worried about VeriSign's Kontiki either, he's "looking into P2P-related opportunities".

Is Mirror Image interested in partnering with hosting companies? Jim says of course - he's established ongoing relationships with "many best of breed providers". Such as Savvis, now that it's sold its own CDN? Jim says no comment. What about a CDN/web host merger, which Dan Golding from Tier 1 Research is rooting for? Jim says he's "absolutely not shutting any doors".

The first thought that crossed my mind was, maybe GI Partners could acquire Mirror Image and merge it with ThePlanet-EV1Servers? (I'd recommend choosing a name in advance to avoid the possibility of triple hyphenation.) But on second thought, TP-EV1's mostly unmanaged (at least for now) service might not be synergistic with Mirror Image's "Global 2000" user base.

Rackspace, on the other hand, might be a good match. In addition to the Fanatical Support it's known for, the company also seems popular among well-funded video startups. Once merged with Mirror Image, the combined entity could go public at CDN valuations! (Akamai trades at 18x annualized sales, versus Savvis' 2.4x.)

Or maybe Jim should leverage his successful, fast growing partnership with Amazon to offer global content distribution for data stored on S3?

Not that Mirror Image has any difficulty getting business. Jim says the content delivery market is WIDE open; at least 50% of the company's new signups come from first-time CDN users. Don't you wish the situation were the same in hosting?

Rich Miller at Data Center Knowledge reports that (a) Level 3 will acquire Savvis' CDN business for $135 million, and (b) Savvis plans to build 4 new data centers in 2007. The new Atlanta, NYC, DC and Santa Clara facilities will add 180,000 square feet of space, boosting Savvis' total inventory to 1.5 million square feet.

Level 3's Kevin O'Hara says the CDN deal will help the company pursue opportunities related to "rich media applications such as video, Web 2.0, multiplayer online gaming and software as a service".

Savvis' Phil Koen says its new data centers will be "designed to exacting standards for security and reliability, power availability, cooling, network connectivity and environmental controls". The company is "focused on delivering IT infrastructure as a service" and will used the space to provide managed hosting, colocation and virtualized utility computing.

Read side by side, Kevin's and Phil's statements seem to indicate that there's an upcoming fork in the web hosting market? Some providers (such as Level 3) will focus on high traffic, public-facing apps. Others (such as Savvis) will specialize in extending enterprise customers' IT capabilities (ie hosting internal apps for a tightly controlled and highly localized user base). And the two paths are apparently mutually exclusive?

Savvis' shares (blue) are up by 227% this year, BTW, while Level 3 (red) gained 94%. (I love Google Finance!) Will LVLT catch up with SVVS as SaaS gains momentum and online video takes off? More importantly, should other hosting providers worry about choosing sides?

ThePlanet, in particular, has no presence outside of Texas, but new(ish) CEO Doug Erwin expressed online video aspirations in an interview with TheWHIR. The company also maintains a game hosting division. Will its highly localized footprint make it particularly vulnerable to competition from mass market CDNs such as CacheFly?

Peter Rip, a general partner at Crosslink Capital, received pitches from two technology startups last week. The companies were at similar stages, but (A) received a 2x higher valuation than (B).

Peter says the folks from (B) had grand visions of numerous use cases that their technology could enable. Unfortunately, from Peter's perspective, "enable is a value-halving word". Enable means value creation is left up to someone else - who may or may not invest time and energy to build a solution on (B)'s platform. Technology is only 10% of success, no matter how much differentiation it provides; the other 90% comes from subject matter expertise.

I think PhotoBucket is a good example that illustrates Peter's point with respect to the web hosting market. PhotoBuckets gets 80,000 daily signups, while even the largest hosting companies with the most generous ad budgets don't. PhotoBucket's value proposition is simple: it hosts photos. Every web hosting provider I know does as well, not to mention website builders, blogs, shopping carts, databases, 2500 email accounts, free Yahoo ad credits... This extensive feature set *enables* many web presence possibilities - but doesn't directly facilitate any.

Peter says when a technology company doesn't feel comfortable focusing on one specific opportunity, it clearly doesn't know its customers. (You fall under this category if you describe your target market as "developers" or "small business owners".) Don't enable, he says - solve. And don't ask customers to build - show them how to use. He promises this formula will do wonders for your valuation.

Following Peter's advice, how about an ecommerce package for brick and mortar retailers? An online scheduling service for appointments-based service providers (doctors, accountants, financial planners, etc)? A developers' community that combines hosting resources with code search tools, an application directory, etc? Instead of *enabling* these groups, might there be potential in building customized environments that feel just right to each?

1&1 CEO Andreas Gauger is, apparently. According to an article in the December issue of Business 2.0 Magazine (pages 126-132), Andreas says everyone fears Bob: "If he doesn't do anything wrong, nobody in the domain business can touch him." I disagree, and I'll tell you why later in this post.

Bob, who received a Purple Heart as a Marine in Vietnam, got into the domain registration in late 2000. His 4 million customers have registered almost 17 million domain names (or a new domain every 2.4 seconds), more than 2x the next largest registrar. His 2006 revenue of $240 million is 71% more than last year, and triple 2004 levels. His operating cash flow will hit $52 million this year, up 70% from 2005.

As Business 2.0 sees it, GoDaddy's two competitive advantages are its outstanding support (920 customer service reps and counting - none outsourced; all in Arizona) and Bob's exceptional talent for generating publicity. Tucows CEO Elliot Noss was quoted saying Bob played the appeal process for his racy Superbowl ads like a maestro. It took 13 tries to pass ABC's decency standards; business climbed each time he was turned back.

Bob made similar points during his recent presentation at the Web 2.0 Summit. I was disappointed with how he came across. The audience remembered him mostly for his proud association with "tasteless and slightly offensive" ads, which are now defined as "GoDaddy-esque". In other words, his approach is old school: get people's attention and sell them your products. Which is distinctly non-Web 2.0. With each new signup, Linkedin becomes more valuable to each member of its community. But there's no place for the same network effect at GoDaddy.

Back in April, I wondered whether GoDaddy should start a small business social network - mostly out of opportunism. GoDaddy was said to be worth $250 million in IPO rumors; around the same time, FaceBook banked a fresh round of VC funding at a $550 million valuation. Could a sprinkling of Web 2.0 stardust help garner investors' favorable attention?

But I was wrong. Web 2.0-ness is good for much more than just a pre-IPO publicity stunt. Richard Rosenblatt of Demand Media - who is featured in the same issue of Business 2.0 (pages 42-44) - is building a "Domains 2.0" business. He's already attracted $220 million in VC investments (based on a $500 million valuation) and hopes to attain a $2 billion market cap by mid-2008. He recently acquired BulkRegister and eNom.

Whereas Bob is monetizing customers' unused domains by plastering Google ads all over parked pages (generating $12K in daily revenue), Richard's take is:

"If you can make that much doing nothing, what if we added some user publishing tools so that people would come back (to contribute content and participate in discussions on topics related to the unoccupied domain name)? What if we built a platform where we could snap that into as many domains as we wanted? That's when the lightning bolt hit me: You'd have a company that generates its own traffic, generates its own content, and monetizes itself. It would be the perfect lazy-man's media company!"

Like Richard, Bob sees Wall Street as a near-ish future destination. He says in 4 or 5 years, "GoDaddy will be up there with Google and eBay among the leading Internet companies." But take a look at Bob's business model versus Richard's? Which is closer to Google's and eBay's success in leveraging the network effect??

I think Bob needs to heed his own advice. "Get - and stay - out of your comfort zone," he tells Business 2.0 readers. You should do that too, Bob! :)

I had lunch with Hillary and Frank Stiff from Cheval Capital earlier. (Their site seems to be hosted at Hostway.) We reminisced about the good old days, when Hillary brokered digitalNATION's $100 million sale to Verio. The deal was all cash and valued at 14x annual revenue. You don't see numbers like that any more.

Of course, Cheval has completed 19 web hosting M&A transaction this year. And Hillary says she's got a number of additional deals in various stages of closing. Negotiations take time, because 9 out of 10 hosting providers practice shoe box accounting. Cheval has represented sellers whose financial record-keeping consisted of floor to ceiling stacks of receipts and merchant account statements. But they can't estimate their server count and don't know how their churn rate is changing over time. Which makes them sooo much less marketable.

Frank and Hillary joked about starting an Accounting Squad (accountingsquad.com is available!). It'd be just like Best Buy's Geek Squad, but instead of computer fixers in VW Beetles, you'd get server counters in Mini Coopers.

On a more serious note, we talked about the sky-high affiliate commissions that so many companies are paying for new sign-ups. I just checked on Commission Junction; there are 17 shared hosting providers who offer $100+ payouts. Hillary's advice is not to get carried away with one-upping the competition. Instead, you've got to see what your numbers say.

Speaking of which, a few months ago Ted Smith from Peer 1 passed along what he heard from BlueHost's Matt Heaton during a HostingCon panel: there's more to advertising ROI than sign-up rate. You should also monitor average churn stats on customers you've acquired through different ad venues. For instance, if a Google-referred customer costs twice as much as Yahoo! but stays three times longer, you've got yourself a good deal. I'm sure Hillary - not to mention your company's future buyer - would approve of this kind of methodical analysis.

GoDaddy. That seems to be the consensus so far on the web hosting IPO prediction market I set up on a service called Inkling a couple of minutes ago. (I learned about Inkling through Bruno Haid, who recently set up a Enterprise 2.0 prediction market.)

By leveraging participants' collective knowledge, prediction markets have accurately picked Presidential election winners (75% more effectively than traditional polls!) and Oscar nominees. HP, Intel, Google, Microsoft and GE are some of the companies that use internal prediction markets to identify new business opportunities, manage manufacturing capacity, forecast sales and decide what quantities of parts to buy.

So if you'd like to know whether an IPO is in the cards for GoDaddy, for instance, or EV1-ThePlanet, or Rackspace, click here and make some trades. By aggregating multiple traders' activities over time, Inkling might be able to give you some pretty good answers.

Inkling is a free service (which looks to be hosted at Saavis). I am not affiliated with the company. Once you sign up, you'll receive $5,000 in credits, which you can use to bet on the IPO-likelihood of GoDaddy, EV1-ThePlanet, Rackspace, Hostway, Endurance International and Aplus.Net. (Please email me if you think of other companies who belong on this list.) The starting price for all phantom stocks is $50. Prices automatically rise with any purchases and fall whenever there's a sale.

Your $5,000 is good for betting on multiple markets (Will Mac OSX launch before Windows Vista? Will Yahoo! buy Digg? Will Britney and K-Fed last through the end of 2006?). The top trader on Inkling has a net worth of $359,703.63!

At Tier 1 Research's Hosting Transformation Summit back in September, Akamai, Limeline, Netli and Solid State each attested that they enjoy close working relationships with their hosting partners. Tier 1 analyst Dan Golding suggested ongoing cooperation is so important that CDN-web host M&A deals might not be such a bad idea.

So how come Savvis is (supposedly) looking to sell off its CDN business?

While Dan argued that web hosts who don't have a worldwide footprint need to help site owners reach far flung visitors through CDN partnerships, Om Malik says the two businesses are at odds: one makes money when people buy bandwidth, and the other reduces bandwidth usage. Which do you agree with?

By the way, Savvis' CDN business, which has 91 employees and 60 clusters in 29 countries, generates $40 million in sales, and is supposedly priced over $100 million.