I have used this blog to beat up on a number of technology vendors. Sun seems to be a favorite whipping boy. I guess I’m guilty of kicking a company when its down.
But, I do need to say something nice about Sun . . . before I bemoan their recent strategy missteps.
In regards to their new SPARC chip T1/Niagara chip, I want to say WOW! In the battle for chip functionality in some markets, the T1 is something special. Sun has managed to create an 8/32 way mini cluster on silicon. When one looks at the requirements of a web focused server farm, or any server sets that rely on threaded symmetric multi processing (SMP), he T1 fills several subsegments of the clustered server market, and perhaps a few of the virtual machine market.
The problem is it that the T1 only serves these subsegments, and the top end of segments at that. Sun continues to box themselves into a corner by holding the high ground while the invading hordes (commodity chips and servers) chisel away at the base of the hill. Since IT buyers want to build relationships with key vendors, they turn to those who provide both their current need and their long term strategy. Sun is not supplying whole products for all these points . . . just the upper tiers.
Here is Guy’s #1 Golden Rule of markets: Never invest in a technology defined by being the biggest, fattest, or one that optimizes bottlenecks. These markets die because technologies develop that eliminate the advantage currently offered.
Take modems and the internet for example. Many people have created slick little technologies to speed throughput on copper bound telephone lines. And the market responded by installing cable modems and DSL. Improving a mature technology is bound to lose over a new technology, or an unstoppable market force.
In Sun’s case, the unstoppable market force is commodity servers. With AMD and Intel in a speed hiking, cache fattening, price slashing slugfest, the growing IT demand for their commodity will eclipse Cell Processors, T1, and other non-competitors. The market has made its strategic choice and will avoid products that are not in the general mold.
How much time does AMD or Intel need to create their own 8/32 chips? Not much given their current roadmaps. They soon will deliver products that also fulfill the high end, and can be bought from Dell, HP, IBM, and Ma Kettles Corner PC Shoppe and Quilting Emporium.
In Sun’s defense, they are trying the classic marketing maneuver of changing the rules of the game. They are hemorrhaging their software IP in an attempt to create a market around Sun and Sun’s remaining revenue creating products. By offering “open” software across a number of platforms, they hope to equal what has occurred organically in the Open Source and commodity computer markets.
But that is a lot to give up to compete with an entire market movement, and one that ultimately reduces their differentiation to all subsegments of IT. Sun may well become the King of the Hill in open systems servers . . . but will spend all of its time, money, and manpower defending their mountain top.