Here's an interesting problem. Through the dot com era (and slightly before), the hosting market was driven to ravenous acquisitions of companies based on the value of their customers. Companies were able to make acquisitions on the basis of their customer contracts - a fairly simple equation.

Cloud Computing is built on utility platforms. It doesn't actually have a lock-in on contracts. Your customer base is stuck with you only as long as nothing else better comes along, or the pain of moving outweighs other factors.

The rising sun of interoperability threatens to decouple this further. As Tom Bishop said on a recent post to the Cloud Computing Interoperability Forum (CCIF):

... we're going to need to figure out what a "packet of computation" is. In particular, one of the most ingenious elements of the original Internet work was the concept of a "dominant decomposition" or "spanning layer," where everything above that layer was a "client" and everything below that layer was a "provider." They even discuss at length the traditional hour-glass shape often referenced elsewhere, meaning that "providers" only use services provided at the layer interface up, and "client" only implement the standard services, and providers and clients otherwise have no specific knowledge of each other. In this way, port a TCP/IP stack to my cell phone and, ta-da, I can get to web sites that never envisioned being viewed on a tiny screen (okay, we all know there are real limitations in many instances, but you get the idea).

This is a key point, and something that seems to be getting missed: if the client is portable, providers are completely commoditized. It's happened over and over again.

This is a common market problem, and not in itself a problem. When you have a commoditized market, there are plenty of mechanisms to distinguish yourself: mostly through brand positioning and marketing. This means that if interoperability continues to rise, and we do succeed in creating interoperable clouds, which is possible (though most "standards movements" usually split into two camps, as Cloud is definitely starting to do), we're going to accelerate commoditization even further.

I was talking with a potential customer for our Elastic2 Drupal Cloud Service today, and he said something really interesting. He said, "I was at Rackspace, but they were really way too expensive. So when the recession hit, I moved over to Hostway. Then I took your Drupal AMI and put it into the Amazon Cloud, and I was thinking about using that to build my infrastructure." He paused - and here's the kicker - he takes a breath and says, "But you know, I really don't care about deployment. My plan is just to keep switching infrastructure as prices come down and innovation continues. Seems like every time I switch providers, my monthly price goes down an order of magnitude, you know?"

Now, it doesn't take a real genius to put these things together, and thank goodness, because I didn't have one handy. There is a valuable lesson in today's conversation: there is absolutely no client loyalty among at least one Cloud Computing customer. And I'll bet there are plenty more that feel that same way. While the cloud community is running around talking about "Not having to worry about infrastructure," customers are playing the same old price/performance game they've always played, and they absolutely want to be paying less than they are now.

Obviously price isn't the only driver, but if we more towards a clearer separation of applications and their infrastructure, combined with a utility billing model, it's going to make brand just about the only separator for different public clouds. Moving between them will be absolutely simplistic, and price wars could easily cause commoditization of entire providers, and there is absolutely nothing, nothing but brand and a few minutes of time, to keep people from moving to the next provider. Provider will be a word, at least to consumers, that means, "the stuff below the application." And as long as it's reliable and cheap, who cares?

So, if the providers get commoditized, no problem. We know how to deal with that. We'll just roll up smaller companies to create larger companies, or large companies will enter the space through acquisition, as has been the pattern in hosting, social media, and just about every hot tech market we've had in the last 12 years. But there's a problem. Wait, there's a huge problem.

Utility Billing.

Yeah, it's the hottest thing since push, and your electric company has been doing it for years, but it turns out that being able to bill for infrastructure with no lock-in is a considerable problem for acquiring clouds. There are no contracts, there is nothing to keep customers from moving. And there is absolutely, positively every reason for them to do so during a transition if the infrastructure is commoditized. And did I mention that plenty of startups are working on brokering clouds for applications to make this migration insanely easy?

So, does that mean that the rush of rollups isn't going to happen? I think it does. Acquisitions will be much more strategic (for team, technology, or position), quiet, or handled very carefully from a marketing perspective, because the recurring revenue that keeps that business alive isn't locked in: spook the customers, and you basically blew it. Those are definitely not conditions for a rash of rollups like the hosting industry experienced from 1997-2001 (approx), where customers were upset, moving through sometimes 2 or more companies a year, but locked into contracts.

I was just thinking about this earlier tonight, and realized this might be one heck of a big problem. Obviously the cloud market is going to segment, and this is really one of the problems with public clouds, not so-called "private clouds" (private infrastructure or internal; they are different), but it could mean that the overall market is going to behave extremely differently from previous technology bubbles.

Remember, this is perhaps going to be the single largest move of hosting customers (and hosting is a massive market: 10+ billion) since the inception of virtualization, and perhaps bigger because Cloud can represent a replacement for massive multi-tier architectures. And it's not just hosting, it's a plethora of other applications coming from HPC, vertical markets, etc.

Thoughts?