Meat's analysis just does not hold true in the real world. Why? because the technology curve is continuous.
The very argument that IT departments take a longer view of assets is why they don't need the premium machine today. I have worked both as a senior IT executive and also worked as an IT consultant with many of the largest IT groups (including companies like BofA, Citibank, Merril Lynch, USWest, and General Motors). They have a continuous stream of technology transfer and injection. The idea that yesterday's hot machine will need to be upgraded but today's hot machine will not is just a fallacy. Each model needs to include the same upgrade shift points throughout its life. The only difference is the captial cost for the technology injection, and where the injection point occurs.
The 266 will lose utility sooner than the 400, but last year's 133 or 166 will lose sooner yet. and the 400 will lose sooner than tomorrow's 450. So the real question for the intelligent IT manager is to manage the lowest cost of ownership over a 3 or 5 year period (depending on their usage model) by picking appropriate purchases and injection points throughout the life cycle. There are always some users who need or demand the newest hottest thing, but the IT manager has to look at the overall flow of usage.
For example, 486 machines are just now falling out of the back end of the pipe for these big companies. They are being used with Citrix software for 'dedicated' uses (1 or 2 applications). This trend allows the older machine to perform like a 166 pentium by offloading much of the work to a back end server. (An aside - guess who has 65% of the Citrix market? Not Dell - it's CPQ). the 486 machines that go into those apps are replaced by 133 and 166 machines which are now too slow for the 'power users' but still support standard 'productivity' uses (email, word processing). Productivity use dominates the big companies, representing 60% of the installed base. The 133 and 166 class machines are being replaced by 200, 233 and 266 machines for 'power users'. some of these are existing machines purchased for last year's 'super user', but since the bleeding edge purchase represents less than 10% of users at these companies, the mainstream purchases outweigh the high end by about 5 to 1. Just for example, BofA has more than 80,000 PCs in the pool. they purchase about 20,000 new ones per year on a 5 year cost model. In a five year plan, why would getting 6 months ahead on the technology curve with no concurrent business benefit be sensible?
Are people buying 333 and 400 machines? You bet. Is there any reason to think that the big IT purchases will shift forward on the curve because of this new technology? No way. There will be 5 mid-class machines purchased for each high end box. The demand curve is not flat and peaks at technology that is 12 to 18 months old, driven by a number of dynamics that have only marginal relation to hot technology. Doing anything else is just injecting money into the stream at a point where it yields no return. Any IT manager who tried to explain doubling his capital investment because he could. by doing so, provide the mainstream productivity users with capability they will not use would be out of a job faster than you can say ROI.
This type of usage analysis and buying pattern is not one man's opinion based on a limited data sample, but is the standard in the industry and has been for many years. |
