What Clouds Destroy They First Make Cheaper

by: Dana Blankenhorn

Self-proclaimed enterprise computing “experts” are wrong about one very big thing.

Software as a Service (SaaS) is not cloud. A SaaS company like Salesforce.com (NYSE:CRM) may be hosted in a cloud environment, but the environment and the resulting service are not the same thing.

Cloud is a new way of organizing computing resources that makes SaaS more powerful and more competitive with traditional enterprise computing. Cloud helps drive SaaS costs to the floor, if applications are designed for a cloud environment.

But SaaS, by itself, is not cloud. Anyone who tells you that is selling you something.

Cloud capabilities are detailed well in the latest National Defense Authorization Act, specifically in Section 2867. As Jerry Bishop notes at TechWeb's Internet Evolution, the plan calls on DoD's CIO to deliver a plan in April that will reduce the following:

  • Square footage of datacenter floor space
  • Power and cooling utility costs
  • Capital infrastructure costs per megawatt of data storage
  • The number of commercial and DoD-developed applications
  • The number of full-time equivalent staff

This is a good description of what clouds do. They reduce the footprint of the data, and all its attendant costs, including personnel costs. And (this is the important bit) software licensing costs.

This is not just because cloud software may be open source, from Red Hat (NYSE:RHT) or VMWare (NYSE:VMW) or Rackspace (NYSE:RAX) . It's because cloud software isn't licensed under an annual per-server or per-seat contract. Instead, clouds are rented based on time and the specific resources consumed during that time.

Now, if you want to buy your own cloud, from IBM (NYSE:IBM) or anyone else, that's different. Just not as different as your Oracle (NASDAQ:ORCL) salesman may have you believe. Because clouds are not built with expensive servers at all. They're usually built with general-purpose computers, because a bunch of cheap machines can be harnessed for less than one expensive one can be bought.

For 2012, my prediction is that the last huge expense involved in building a cloud, those expensive data storage units from HP (NYSE:HPQ), IBM or EMC (EMC), will start to go away. This is partly due to Red Hat's integration of Gluster into its CloudShift offering. Once data storage and delivery costs are reduced, so that general purpose drives and drive networks can replace single-purpose systems, cloud economics get another gear. (This is also the only cloud on the horizon for EMC, which owns 80% of VMWare.)

Another big story for 2012 was announced this week by Amazon (NASDAQ:AMZN). That is, the scaling of cloud capabilities to supercomputer status, and the delivery of that capability as a standard cloud service. Another big boost for the services market, and another reason not to be buying big iron.

In short, 2012 will be the year many cloud companies justify their high price-earnings multiples, and it could be the year when traditional enterprise companies start to feel some real pain. Unless these cloud guys get taken out soon, the big boys won't have the capital advantage needed to buy them up.

Something to speculate on as the calendar turns.

Disclosure: I am long GOOG, IBM.