You are in: Design | EDA and IP

WORK BREAK

Read The Magazine

Issue: 16 - 22 Dec, 2009

Warren Savage On: The Shifting Sands of Semiconductor

Monday 26 October 2009 10:57

I've been associated with the semiconductor industry in one form or another for about 30 years. During that time I have seen a number of boom and bust cycles, the rise of the System-on-Chip era and the maturation of the IP business model. I think we are at the start of a new era as the semiconductor industry emerges from the current recession.

One of the great downsides to Moore's law is that it is increasingly difficult to put to use all those new transistors that are given to us every 18 months. And worse yet, its more and more expensive to make the products that we inevitably get to sell for the same price (if we are lucky). Something's got to give.

If you look at the three elements of the semiconductor industry - silicon, EDA, and IP - you can see each part of the industry adapting in unique ways that we haven't seen so far.

On the silicon side, we see the giants of semiconductor shedding product lines and businesses. In Europe for example, ST Microelectronics, Ericsson, NXP, and Infineon are selling off business lines, forming joint ventures, and focusing on those businesses that generate the highest profits and abandoning low margin markets to competitors who are focused on winning on price.

In doing so, they stop playing the race-to-the-bottom game of the last 10 years and begin an era where their products receive more value for their underlying intellectual property and know-how instead of simple cost-plus calculations based on silicon area. In other words, semiconductor pricing will stop tracking Moore's law and be driven more by the value of the content of those chips to the customer's application.

On the EDA side of things, let's face it - it's been a mess out there in recent years. The EDA industry built itself on a Microsoft-like business model of licensing their tools to users (i.e. engineers) and has also found itself on the wrong side of Moore's law.

More transistors mean larger chips, and larger chips means fewer projects and therefore fewer engineers who need access to the tools.

Meanwhile, EDA R&D costs are up because they need to support the latest processes to keep pace with the foundries. With a shrinking market, EDA have found themselves in a zero sum game of stealing market share from each other. Instead of narrowing their focus like their semiconductor customers, EDA companies are emerging from this recession through the strategy of expanding the capabilities beyond traditional EDA. They are moving into new business areas like IP, prototyping, and software.

Synopsys, for example, is the number one company in EDA, and has now zoomed to the number 2 position in IP. The lesson that seems obvious here is that when your main business is drying up you should look for new ones.

Finally, let's talk about the IP business. Thanks to Moore's Law, IP has been growing at a much faster rate than both EDA and semiconductor for the last 10 years. That's because the more transistors that are used on a chip, the more likely that those transistors will be made of licensed IP. The economics of reusing IP from previous designs has trumped any claims for design productivity through better EDA tools and cheaper engineers from the east.

IP could be the most important and profitable segment of the semiconductor industry as we move into the next era of semiconductors. Already we are seeing substantial acquisitions and consolidations in the industry like Virage Logic's acquisition of ARC and Synopsys' acquisition of Chip Idea from MIPS. As these companies disappear into their new parents, new IP companies will emerge to fill the void and the cycle will repeat.

I think one of the most interesting dramas playing out in semiconductor at the moment is the looming battle for the embedded processor market between ARM and Intel. That a relatively small company from the sleepy tweed hamlet of Cambridge can give fits to the most powerful semiconductor company in the world is a true testimony to the power of the IP business model and changing dynamics in the semiconductor industry.

Reinvention is essential to the long-term survival of any business and it seems that the semiconductor industry is well on its way towards its own renaissance. Most people don't remember that IBM used to be famous for typewriters. Perhaps we will all have stories for our children someday that Intel used to be famous for PC chipsets.

Warren Savage, President and CEO ofIPextreme, is a well-known and published authority in the field of semiconductorintellectual property.

He has a long history of pushing the envelope of design methodologyfrom his work in fault tolerant computing at Tandem Computers inthe 1980's and driving reliable design methodologies intocommercial practice at Synopsys for its DesignWare IP product inthe 1990s. Much of his thinking became embodied in the seminal bookon IP reuse, the Reuse Methodology Manual. Catch him on Twitter atwarren_savage andipextreme

Previous columns

(Nov 07)Warren Savage On: Making the Case for Invented Here

(Dec 07) Warren Savage On: Swiss Cheese Solutions

(Jan 08)Warren Savage On: Collaboration Needed for Success

(Feb 08) Warren Savage On: Knowing Your No

(Mar 08) Warren Savage On: The Next Big Thing

(Apr 08) Warren Savage On: Gumming Up the Works?

(May 08) Warren Savage On: Waiting for Godot

(Jun 08) Warren Savage On: Our Virtual Future

(Jul 08) Warren Savage On: Being Plugged In

(Aug 08) Warren Savage On: The Dog Days of Summer

(Sep 08)Warren Savage On: Samurais, Sedans, and Semiconductors