Itanium or Itanic?

It is a foregone conclusion that the Itanium processor will be a commercial

flop in terms of return on investment. A decade of development involving

hundreds of top-notch engineers and costly launch delays have put a mega-big,

make that a giga-big price tag on this product. Intel will never recover that

investment under the best of scenarios.

To top it all, the worst thing that can happen to any new product is

customers knowing that there is a better version right around the corner. But

that is exactly the position the three-year launch delay has put Itanium in, as

the next-generation Itanium, the McKinley is only 12-18 months away from

production.

Itanium will not have a fair chance to enjoy a typical product life cycle. It

will instead function largely as a test case, a beta, for McKinley. Many

companies will buy a few systems, but most network administrators will likely

wait for McKinley to place more significant orders.

Another factor hampering Itanium's success is that the release comes at a

time when the market, especially for servers, is struggling to find new footing

after the bottom gave way late last year. Even Intel is cutting back on the

plans for its server farms, where many Itaniums would likely have ended up had

they been available.

And just because it is here and it is from Intel, does not guarantee success

in the server chip business. A decade or so ago, I attended the launch of the

Intel i960 series of risc server processors and the demos that showed i960

machines running circles around similar Sun Sparc workstations. But the i960

flopped and there is no guarantee that the Itanium won't do that either,

especially if software developers prove slow in making applications available

that take advantage of the 64-bit architecture in some kind of compelling way.

A final potential trouble spot is the perceptually "slow" 733 to

800 MHz at which the chip is being introduced. It seems slow when compared to

1.7GHz Pentium 4. One would think that Intel would commit its top-on-the-line

production techniques for server chips that need to compete with the UltraSparcs

of the world. Not putting its best production capabilities to work for the

expensive Itanium, seems a poor business decision.

Still, May 29, 2001 is important because the Itanium is at last available and

a new chapter in the server computer market has begun. No doubt it will fill

many pages in the years ahead.