What’s Next? Semiconductor Leaders Speak, Part 1

Throughout history, the semiconductor industry has continued to change and evolve — and we have looked to industry pundits to point the way. That conversation is a critical part of forging our new path.

Recently, I read a volume titled Fabless: The Transformation of the Semiconductor Industry, by author Daniel Nenni and contributor Paul McLellan. In the final chapter, the authors polled a variety of industry names with the question: “What's next for the semiconductor industry?” I wanted to engage with these thought leaders with some ideas of my own, taking thoughts I've developed and written about on macro-economic principals based on the theory of mass capitalism.

In today's first virtual sit-down (first in a series), I'll be addressing the thoughts of Simon Segars, chief executive officer of ARM; Aart de Geus, chairman and chief executive officer of Synopsys; Lip-Bu Tan, chairman and chief executive officer of Cadence Design Systems; and Dr. Ajoy Bose, chief executive officer, at Atrenta.

Segars makes in interesting statement in the book, when he says, “Without the seamless design-to-end-user supply chain that has evolved through the growth in the fabless industry, our vision would remain a dream.” He believes that the movement toward fabless semiconductor companies has democratized innovation.

I am in agreement, but would add that the growth of small businesses should be credited with the success of the fabless semiconductor industry. We are still striving for democratized innovation, which will only be achieved when we have a functioning economic democracy.

Meanwhile, de Geus puts forth that semiconductor innovation would be most visible through utilization. I believe that this utilization must be of a progressive nature for innovations to benefit the masses. Only by means of benefiting all could it be possible to visualize true benefits of innovation. For this to materialize, a free market economy has to be established.

He also points to the enablement of “better” and “sooner” innovation (rather than simply cheaper) as a desirable means of keeping pace with the speed of innovation in the semiconductor food chain. I would respond that the best, and perhaps only, path to this sort of innovation is a decentralized supply chain.

Cadence's Tan calls for deep collaboration in order to reach higher levels of innovation and financial success in the supply chain. He says that no one in this ecosystem can succeed alone, but we can all succeed together.

This collaboration can be achieved only when all participants in this fabless semiconductor ecosystem have a stake in the success of the entire ecosystem. Hence, there has to be a change in corporate business managerial systems to ensure accountability and success are shared with all participants, which includes every employee that is a part of the fabless semiconductor ecosystem.

Finally, Dr. Bose of Atrenta suggests that low-margin demand may be a simple reality in the semiconductor market, the only possible exceptions those elite few able to move up the chain to offer more total ownership of systems and services. He further says that there would be a continued drive to move design to low-cost regions, and the clock is running out for any suppliers unwilling to adapt.

Bose is right. The cause for low-margin demand is lost consumer purchasing power due to US macro-economic policies. The Fed's monetary policies since 1970 have resulted in Crony Capitalism, which has transformed the US economy from a free-market enterprise system to monopoly capitalism. This has resulted in total ownership by an elite few who. Following policies of free trade globalization, which drive down the costs of design by moving design operations to low-cost regions, would further exacerbate the problem of low marginal demand because of trade deficits that would be added to the US economy.

Let us know how you would respond to these semiconductor industry pundits in the comments section below.

10 comments on “What’s Next? Semiconductor Leaders Speak, Part 1

  1. apek
    July 9, 2014

    No. In fabless foundry model, both are partners and both collaborate and benefit from the success. It is based on mutual trust, cooperation and collaboration…fabless does not have to invest in capital intensive fab and foundry can implement designs from several fabless businesses to reduce its idle time

  2. _hm
    July 10, 2014

    Fabless semiconductor industry is small part of total industry. It gives diffrernt avenues to new innovation. But to be leader, one needs its own fab.

  3. SP
    July 10, 2014

    Unless you have production numbers like Apple and Samsung, it doesnt make sense to own a fab. Its not an easy business to manage. Although its attractive from complete ownsership point of view the risks to survive are just too many.

  4. apek
    July 10, 2014

    You believe it or not…but fabless is future of global semiconductor industry for sustaining Moore's law

  5. Rodney Brown
    July 10, 2014

    Apek, I am interested in hearing how fabless is going to sustain Moore's Law, because isn't fabless just another word for outsourcing or offshoring? I mean, somebody still has to have a semiconductor foundry, to make the chips designed by fabless semiconductor firms, yes?

  6. apek
    July 10, 2014

    Rodney, Present fabless model needs some tuning…I do not believe in model based on outsourcing. Please look forward for future parts in this series.

  7. Rodney Brown
    July 10, 2014

    OK, I can be patient. 🙂

  8. apek
    July 11, 2014

    Well…not necessarily. I think we have lot of faBS sitting idle in America and this does point to the failure of IDM model

  9. apek
    July 11, 2014

    Public faBS have to be managed as efficiently as private faBS. This would give dual advantage of growth of small fabless players and sustainability of capital intensive faBS

  10. apek
    July 11, 2014

    Meanwhile please read synopsis of my upcoming book. You can view it here

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.