I spoke with OpSource CEO, Treb Ryan today. He's the first person I've talked to in the cloud hosting business who has spoken candidly about the actual profitability of being a cloud provider. According to Ryan-who's company is a relative newcomer to the cloud business-the margins are "the best I've seen in the hosting business." He called them healthy, after smiling and explaining just how simply the math comes together on the whole model.

Of course, Amazon has made it sound like the margins on cloud hosting are very thin. Clearly they are sandbagging. Or they are doing it completely wrong. Which do you think is more likely?

Me too. Cat's out of the bag, Amazon. Cloud hosting is ludicrously profitable. Here's why:

The math works out

Here's where Ryan made his best case: 1TB of disk space at Fry's is $100. From NetApp, it's around $700. From cloud, it's about $1800 per year. Any ops manager can do the math for ROI on that equation. Ryan says this actually plays out in the end business of cloud hosting. EDIT: To clarify, obviously running a data center costs more than the disks. Also obviously, no cloud provider would use disks from Fry's. That's why I quoted a price for NetApp disk arrays per TB. The point is that once you are running a data center, offering a cloud is going to make you more money than offering traditional hosting. Yes, there is cooling, admin, support, and tons of other stuff you have to have in order to offer a cloud. But considering how much you can charge for cloud servers, the numbers still work out tomake cloud hosting a profitable business.

Customers are lazy

Obviously, Ryan did not say this. But its still inherent in the math. If the customer screws up, they're likely erring in the favor of the provider. Sure, 4 cores is needed at 4 in the afternoon, but is it needed at 3 AM? If not, you're still paying the same rate to keep that unused CPU time up and running.

I'm sure most of you reading this have a cell phone plan, and have, at least once, had a minutes overage. Your phone provider makes a great deal of money from that type of contractual payment plan, and the same principle applies to cloud hosting.

This relates to the previous reason: customers are lazy. If your application calls the database like a gossipy teenager on a Friday night, the cloud provider is going to win out. We've all heard the story of the nightmare application that calls the database 10 millions times an hour to add 2+2. In the past, silliness like this has resulted in more database licenses sold by Oracle or IBM or Microsoft. More recently, it meant service contracts to MySQL houses and a new rack of Dell servers. But now, it means more CPU time in a cloud. Which means more billable time to the cloud provider.

Instead of paying company X, you're paying the cloud provider. 

I hate to be the negative nancy, but we all know there is a lot of terrible code out there? The code hosted in these clouds doesn't have to be all bad to make the numbers work. A healthy 10% would be enough, provided those overages in usage were as gross as those of legend.

The moral of this story? Cloud hosting is a very profitable business! And the way to save money, as a customer, is to optimize your code. For projects hosted in the cloud, you can actually get a monthly statement that can effectively reflect your performance optimization efforts. Change the code to hit the database less, save a few hundred dollars per month. The cloud can give you a direct line from your code to your dollars. Heavy stuff!

I have to confirm the number Ryan spouted before I print it here, but the percentage gross he quoted me was too good to be true. Ryan is speaking as the keynote at 8:30 AM tomorrow at the All About The Cloud summit. It should be the most interesting presentation of the show, which was otherwise dominated by vendor talks.