Srinivas Bharadwaj's Weblog

Trends in the server market - more on the M. Effect and Herfindahl Index

I must first state that it might contain forward-looking statements and that I own Sun stock and I am an employee and I am not an officer of the company. This is not a formal analysis by any means and must be viewed merely as a hypothetical musing. No significant work has been done on this front.

The Average Selling Price of Dell and "the rest" in the server market is somewhere around $3300 a server. Dell and the rest own approximately 25% of the market.

The commodity server market is 7 million units strong. And it is a $25 billion industry. It is also an industry which is seeing increased volumes and revenues coupled with decreasing per unit contribution margins.

While Dell's costs are slightly below that of "the rest" (due to economies of scale, SG&A amortization and efficiency through co-selling PCs, etc), the commodity market is a tight place to be, given the supplier power. HP is hot on Dell's heels.

It is possible to calculate a separate Herfindahl index for the commodity server market. The important point though is that the Herfindahl index has trended higher in this market. Yet it can be said, that the level of competition indicates that there is at best single digit operating margin available ( under $300 a server). The reason is again supplier power. But it is important to understand that it is also supplier power that keeps the Herfindahl Index higher. For instance, had there been dedicated MIPS, a Power PC and a Sparc chip maker, all going after the same server market(this could also have happened,- yes, really), the Herfindahl Index could have been similar to the PC industry.

Perhaps this is the reason why Apple has avoided the commodity server market.

For companies like Sun, HP, IBM and Fujitsu/Siemens, this is a more complex equation.

If one were to trend the Herfindahl index versus time(measure Delta H/ Delta t) and compare this with the Delta(operating margin)/Delta time, there is likely to be significant correlation. But it is likely non-linear.

The commoditization of servers has been a central theme in the shaping of the Internet. As the Internet starts to include more media and IPTV, more virtual reality, as it goes more increasingly mobile, - creative disruption can alter the dynamics of the changes in the value chain, and additional richness in its requirements could become more significant.

And this creative disruption is what can change the symptoms associated with the Molodovsky effect. But the current state of affairs is yet to reveal any such disruption, barring the utilitarian direction, be it Search or SaaS. As services infrastructure evolves, so too will the role of the companies and the value chain, and this will alter market thinking and perceptions on growth, and the idea of the role of M. effect in server markets.

Posted at 02:41PM Jun 12, 2007 by participationist in Finance  |