The Texas Surprise

 Gary Smith

The Texas Surprise

The consolidation of the semiconductor industry has been a hot topic for at least five years. Although this consolidation has usually referred to the elimination of the IDM model, it actually is the consolidation of those semiconductor companies that fabricate silicon products. Today that means IDMs and foundry vendors. In the past, Texas Instruments (TI) always made the list of IDM survivors. Even with problems with the Crolles2 Alliance, STMicroelectronics usually made the IDM list. In ST’s case it was expected that it would acquire one or more of the other semiconductor IDMs; often Motorola/Freescale was mentioned as a possible acquisition candidate. The combination of cell phone and automotive emphasis made that union very attractive. So the recent collapse of the Crolles2 Alliance and the subsequent announcement that TI would stop R&D for the 32 nm silicon node process came as a surprise. We now need to look at this “consolidation” more carefully.
Defining Consolidation
First we need to define what we are talking about; too often this issue has been sensationalized for public consumption. By the turn of the century it was obvious that the semiconductor market was splitting into two camps. In one camp were the traditional semiconductor component suppliers such as ON Semiconductor, Fairchild Semiconductor and National Semiconductor. In the second camp were system IC vendors, such as IBM, Intel, AMD and STMicroelectronics. Although memory is obviously in the component camp, its requirements for leading edge silicon and significant manufacturing capability place those product vendors in a separate category. The consolidation we’ve been talking about is the consolidation of the system IC vendors that primarily supply the logic and mixed-signal ICs that are a significant part of a system’s design. These are the leading-edge vendors for whom performance, and usually price, demands the most advanced silicon in high volume.
A Little History
The history of this consolidation is interesting. During the 1980s semiconductor process R&D was difficult. We shifted from the bipolar process to NMOS and then to CMOS as the silicon process of choice. During this time we developed many other processes that either faded away or became niche markets. By the 1990s we had settled on CMOS and entered a period we call the “Cream Puff” era of semiconductors, owing to the relative ease of IC creation during that period. Process development was incremental and fairly obvious. The research effort was primarily focused on semiconductor equipment, with UV being the great hope for the deep submicron nodes. This was a period where silicon processes were considered, by some, a commodity and the foundry vendors became a vital part of the semiconductor design chain. With the emergence of foundries also came the fabless semiconductor companies and the semiconductor market evolved into what it is today.
130 nm and Beyond
The 130 nm node changed all of this. The introduction of copper interconnect took a significant amount of research and we started to see IBM and Intel rise above the other semiconductor companies. Unfortunately we had gotten into the habit of thinking of R&D as one process, when it actually is two distinct activities. All of a sudden there was rampant discussion about how the cost of process R&D had gone through the roof. That isn’t exactly what happened, though. The cost of process development had increased, but not enough to drive IDMs into the fabless world. Looking at the list of component vendors, many are not expected to go fabless. Their process development budgets are within their capability to continue building fabs. Of course these are not the five billion dollar fabs that receive all the press coverage; these are lower-cost fabs utilizing older process technologies. But 8-inch fabs are fully capable of keeping component IC vendors competitive.
The real cost spike has been in the research side of the equation, which is handled mostly by the Intels and IBMs of the world. Once this research has been published in academic journals, the semiconductor equipment vendors are more than willing to help make the process technology available to the rest of the industry. Add to that the ability to buy older fabs from the IDMs that are going “fab lite,” or buy used semiconductor fabrication equipment on the very active used equipment market, and the component IC vendors will not have much problem continuing to operate as usual despite the high cost of process research. So the consolidation phenomenon really centers on the system IC vendors and it is being driven by the cost of silicon process “research.”
Gary Smith EDA
Conclusion: In order to address the so called Semiconductor Consolidation it is neccesary to understand exactly what is happening, what is driving the consolidation, and who is being affected.
Views: 3350
Domain: Electronics
Category: Semiconductors
weSRCH App on Apple

Why Hire Emergency Locksmith Services

Emergency locksmiths have great skill in offering assistance during unforeseen situations of lockouts and lock problems. Read this presentation to know about the significance of em

Gary Smith
28 July, 2016
Gary Smith
02 March, 2016

Locksmith in St. Louis MO - What They offer

Locked yourself inside a car or house accidentally? Hire Locksmith in St. Louis MO, as they offer variety of services at affordable price. They will unlock your car and house in no

Gary Smith
02 January, 2015