terremark-cloudeconomics

Cloud Economics, By The Square Foot

4 comments

A chart from Terremark Worldwide compraing the costs and revenues for its colocation, managed hosting and cloud computing operations.

As data center and hosting companies transition to a cloud computing model, we’ve been tracking the economics of cloud business models for providers. Cloud computing may be the most economic way for end users to deploy IT capacity. But what about providers?

Is cloud computing more profitable than dedicated hosting or managed hosting? New data from Terremark provides additional insight into this question.

Much of our data on cloud economics has come from Rackspace, whose disclosures allow an analysis of cloud operations by average revenue per user and revenue per server (which Rackspace sees as the best metric).

Analysis of Square Footage
In a recent presentation to analysts, Terremark (TMRK) looked at its colocation, managed hosting and cloud computing operations based on how these services used each square foot of data center space.

The result? All three services had the same baseline cost of $1,200 per square foot for data center infrastructure. The managed virtualized hosting and cloud computing operations have an additional cost of $6,800 per square foot for the racks and servers to run the applications.

Strong Revenue for Managed Hosting
On the revenue side, the managed hosting space generates $20,000 a year in revenue for each square foot, compared to $12,000 a year for cloud computing and $1,000 a year for colocation. That works out to an annual return of $2.50 for every $1 invested in managed hosting, compared to $1.50 for cloud computing and 90 cents for colocation.

While colocation offers the lowest cost of entry, it also has the lowest return. Cloud computing and managed hosting offer a healthy return on investment, but also require an investment of $8,000 per square foot (at least in Terremark’s operations; others may do this for less).

That’s why we won’t see the landscape overrun by cloud computing data centers. Cloud economics offer faster, cheaper deployment for end users, but the barrier to entry remains high on the back end. And the $8,000 a square foot investment in infrastructure only pays off quickly if you can fill the space with customers. For many providers eyeing the cloud, a strong track record in acquiring customers is a prerequisite to access to capital.

Efficiency As a Cloud Differentiator
Some cloud providers can lower ther costs by repurposing servers from its managed hosting operation, extending their live by incorporating them into a cloud cluster (see Is Your Cloud New or Pre-Owned? for more). This is key to the benefits of the cloud model for Rackspace, whose cloud operation has a much lower average revenue per user (ARPU) than its managed hosting business.

But Rackspace says that the efficiency of cloud computing, and the ability to pack many users on each physical server, makes cloud computing a winner in a revenue per server analysis. The ability to automate tasks that previously required support staff is another key benefit of the cloud model. Lower supports costs don’t appear to be factored into the Terremark data, and certainly would narrow the profitability gap between cloud and managed hosting.

The higher return on capital for managed hosting is among the reasons why this hosting model will never be fully superseded by cloud computing. Many customers will continue to pay a premium for high-touch support and service for their mission-critical IT assets.

But cloud computing offers a middle path, offering cost and usability advantages for customers, as well as an attractive return for providers. For established data center builders and operators, this creates a “best of both worlds” scenario: a fast-growing customer base and a high barrier to entry for new players.

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)

4 Comments

  1. My training in the Dismal Science comes entirely from the School of Hard Knocks, so perhaps I'm a tad off-base here but… as somebody who has been inside the belly of this particular beast for a long time I note what is likely the largest cost(s) of all missing from this calculation: People. The more "service" oriented a provider is the more people they will require. People add tremendous costs to any operation and without their costs factored in the ROI time can not be calculated properly. Managed hosting and Cloud deployments on the provider side are going to require far more human beings to create, develop, and tend to them than your average colo. All that "automation" that you refer to has to be created and maintained. You'll need software developers, systems administrators, network engineers, and technical support staff, etc. This is just to support the systems alone, I'm not even getting into sales, marketing, administration, etc. Since the creation of software fro automation, scaling, deployment etc. has to be done prior to installing the first customer of either a hosting or cloud deployment it should be thought of, and accounted for as a capital expenditure. By comparison the colo provider can get by with literally a skeleton crew. Of course the more towards the "managed" side (and any colo that is well managed for that matter) will have more than just the basic staff, but I know of plenty of rather large (>15,000sq') colocation providers who employ a handful of human beings. Now let's run those numbers again and see how it looks.

  2. Part of the dynamics is also pricing. Managed hosting can generate more revenue because it is typically priced at a higher level than cloud. Cloud uses a usage based pricing model. And since most sites have low usage levels, they don't generate a lot of revenue. Managed hosting typically locks companies in at a cost level that includes more resources than they will use. And of course colo can generate a ton of revenue if you get a customer who purchases a lot of bandwidth. The nice thing about cloud is that it scales up well, so for providers that can get a good volume of customers, it is a great thing to offer.