It's nice that the author of this artice explores the tangible costs associated with the decision to own or rent server space. I didn't see any of the intangibles or risks!
I love it when I get mail from a legal firm stating that my identity may have been lost due to IBM losing a shipment of backup tapes, or when I receive notes from a Credit Card company stating similar circumstances.
The point is, companies are out to generate revenue, and to protect intellectual property. Imagine if an employee from a "Cloud Provider" decided to sell private company information to a rival for personal gain (Please don't say we are protected 100% from this). The costs associated with this type of activity are not accounted for in your estimations / hypotheses above.
What if we calculated lost productivity due to network banwidth, outages that cause employees to have to work overtime etc...? The list goes on and on. Cloud computing is a nice dream, but when the "Bean Counters" start adding up the costs associated with the obstacles listed above, I am sure the decison landscape changes dramatically.

Two comments:

First, to the point of the financials, another consideration is the time value of money. A lot of IT folks don't take the time to understand the financial aspects of their business cases. For example, money spent today on CAPEX is not the same as money spent over a 36 month life (like lease vs buy analysis). Discounted cash flow investment models take this into account and the economics swing even faster into the domain of "rent" - at least for most companies.

Second, IT executives can find lots of reasons not to do something...especially when they couch them as "Yeah, but what about..." Intangible issues need to be addressed (like risk), but I cringe when it sounds like excuses not to give a promising technology a chance when the unspoken underlying reason is about control. Manage risks, but let go of control...your business will ultimately show you who's really in control.

Great feedback above! As an IT professional (with the required Academic and Practical credentials) with a strong background ranging from development to project management (MES, ERP, Data Warehousing, and Warehouse Management Systems) I find it very alarming when companies are advised to do as CTODave suggests above: "Let go of the control, and your business will ultimately show you who's really in control!"
Dave... it's not the Finance department! From an operational perspective, IT has become just as important as your R+D staff. All of your companies secrets are held in your systems. Internally, you can take measures to guarantee the safety of your data; with Cloud computing and allowing others to house / control your data you cannot (Guarantee it that is). You can hope that your vendors respect your contract, but you cannot achieve the same level of control as internally owned and operated systems.
More than likely we will never reach an agreement on this topic. So, I will leave you with the following thought. I am all for new technology (After all, I am and always will be a software architect at heart), managing risks, exploring new options and maintaining excellence for my organization, but cloud computing will not live up to it's hype. It only takes one violation of privacy to kill the entire concept.

I agree that simple-minded approach is insufficient. And I would also add “air”, often “precision air” to your list of additional direct costs. However, I am afraid that we might as well “complicate” your approach a bit.

Here is why:

1. Why are we using 1U in our calculations and not 1/2 U?

2. Companies often lease the equipment, which skews all calculations.

3. If we are talking about capex vs. opex, why don’t we also talk about fixed vs. variable costs? So, unless the entire data centre is moved to the cloud (which IMO would not be a right thing to do for a few more years, if ever), then the infrastructure and IT Ops expenses are there to stay since these are your fixed cost.

4. I am afraid that any generic calculation would be too generic to make sense to any CIO since the economics of any organization is highly specific.

So, I am not convinced that we can say upfront which way is less expensive for a particular case.

I agree that economics is only one of the aspects of the topic, and I am thankful for your series of articles. However, overall I am not convinced that there is an easy way to slash IP Ops expenses today, in March of 2009. Maybe one day…

Otherwise, I agree with your closing paragraph.

Truly,

Eugene Nizker,
Evident Point Software

Loved your book on Virtualization!

As with Virtualization, the Cloud has a set of contractual and security risks that need to be understood and resolved before some applications are put there - software licensing, change and release controls, etc. - that can wipe out any savings that might be otherwise achieved.

There would also seem to be an opportunity for insurance on the risks of data privacy violations and the resulting costs.

Organizations that achieve mature IT Asset Management / Software Asset Management policies, processes and execution, integrated with Change Management and Release, are more likely to yield the financial outcomes that they hope to achieve.

The real issue is that both OpEx and CapEx can have both fixed and variable cost elements. When one defines profit as Free Cash Flow, and includes both OE and CE in the economic equation, you get some new relationships in a profit maximizing equation for price.
Price = dFCF/dQ - VarOpEx - VarCapEx

From the engineer's viewpoint, it does not matter whether its OE or CE, but both need to be included in the equation for optimizing profit.

Sorry, I specialize in this field and am more familiar with railroad economics, part of my dissertation work years ago.

Avery Grimes, P.E., Ph.D