Analysts tend to make judgments regarding Intel based on an existing conventional wisdom (CW) and projecting straight line into the future. As a former Intel, Cyrix, and Transmeta processor marketing guy I would like to offer a different perspective as I have been both inside the tent looking out and outside looking in.
The current CW is that Intel is doomed… it’s OK, we have been here before. Each time CW says Intel is doomed they implement what I will call their Barbed Wire Strategy to counter their threats and expand their market and influence. If I may, I will explain the Barbed Wire Strategy.
A month ago I moved with my Family to Austin. I had the job of driving our car with one of my boys from Silicon Valley through Arizona, New Mexico and West Texas. In West Texas there are a lot of ranchers with big tracts of land (thousands of acres) all ringed with barbed wire. It is not particularly high but it is there to keep cattle in and people out. These days it’s also keeping in a lot of windmills. Typically the ranch house is deep inside the property off of a long dirt road that most people couldn’t find the entrance to. Consider the ranch house like Intel’s processors. They are very valuable and have existed forever. If someone were to invade the property they would not make it to the house. Now if the rancher wants to increase his property to handle more cattle (or windmills) then he can buy the property next to him and move the barbed wired fence farther out and increase his personal wealth.
Today’s CW is that the ARM Camp has Intel’s number and the game is over. My take is that Intel is already well down the path to implementing the Barbed Wire Strategy on a number of fronts. I will talk about servers today.
Warren Buffet talks about investing in companies with high castle walls and big moats however, in the ever changing tech business you need the Barbed Wire Strategy. At the first sign of a competitive threat – Intel looks to expand the property lines and move its Barbed Wire farther out from the center of the ranch. This past week they acquired Fulcrum– a switch chip startup that competes with Broadcom. Except no system guy would buy from them because of the fear they would not be around in a crunch. Intel acquired Fulcrum in order to own the whole line card in the Data Center (sans DRAM). The switch business is around $1B for Broadcom. So as Broadcom commoditized Cisco’s switch business the past 5 years, now Intel will commoditize Broadcom and Marvell’s switch business. Intel may not get the whole $1B of revenue but it will be additive and more importantly it will move the barbed wire fence farther out from the ranch house.
For someone new to understand this you need to review history. The clearest example I can give is back in the early 1990’s, Intel was facing a resurgent AMD and new processor vendors Cyrix, Nexgen and C&T. The chipset market was a thriving 3rd party market. Intel wanted to increase the barriers to entry for all so they took it upon themselves to develop their own chipset for Pentium. The chipset added a minor amount of revenue, but the protective barrier it set up allowed Pentium prices to rise dramatically. Chipset vendors melted away along with Cyrix. AMD acquired Nexgen.
Expect Data Center revenues to rise with the Fulcrum acquisition and more importantly start thinking about the impact this will have on the ARM vendors aiming for the server space.
More Barbed Wired Stories to Follow.
Note: You must log in to read/write comments