Texas Instruments and the Big Chip Maker Anachronism

Mark is a member of The Motley Fool Blog Network -- entries represent the personal opinion of the blogger and are not formally edited.

OMAP 5 Fails in the Mobile Market Place

On Nov. 14, Texas Instruments (NASDAQ:TXN) announced that it was pulling out of the mobile device (phones and tablets) systems-on-chip (SoC) market and cutting 1700 jobs.  TI’s principal mobile device products, the OMAP line of SoCs will be redirected to embedded applications such as automotive and robotics. TI management believe they can compete effectively in these areas due to longer product life cycles and lower required investment.  Texas Instruments also cited the fact that large customers “are increasingly developing their own custom chips.”  

Given the explosion in the mobile device market and generally positive reviews for the yet to be released OMAP 5 mobile processor line, the withdrawal from the mobile device market might seem ill advised, but in fact TI had little choice.  The Wireless division, which also is home to the OMAP processor family, has been hemorrhaging money, losing a total of $129 million for 2012 through September.  Revenue for the first nine months also steeply declined, from $1.8 billion to $1.04 billion.  TI stated that most of this decline was in their “baseband” products (used to connect to cellular networks) and that the decline was partially offset by a revenue increase due to OMAP sales.  Nevertheless, it’s clear that revenue from OMAP hasn’t kept up with the cost of bringing new OMAP processor generations into production, and the OMAP 5 has failed to bring in new high volume customers.

Emulating Intel

To a large extent, OMAP has been an attempt to emulate the Intel (NASDAQ: INTC) CPU formula: build the best processor you can, and the device manufacturers will come.  Like Intel, Texas Instruments is vertically integrated in that it designs and fabricates the ICs it sells to OEMs, which put them into consumer devices.  Like Intel, TI hasn’t tried to get into device design, although they do provide “reference designs” for the OEMs to use.  And like Intel, Texas Istruments sells mostly a standard catalog of ICs, rather than custom designs.  

The Intel model hasn’t worked for Texas Instruments for a number of reasons:

  1. Agility, or the lack thereof.  TI first announced the OMAP 5 series in February 2011, tantalizing the mobile device world with the prospect of a multi-core ARM Cortex-A15 processor.  A year later, TI displayed a working demo device at Mobile World Congress in Barcelona.  Currently the chip is “sampling,” meaning that samples are available to manufacturers to test and develop with, but not in production.  Production is not expected to start until early next year.  Meanwhile competitors such as Samsung began sampling their own A15 processor about a year ago and are now readying them for production.  
  2. The ARM (NASDAQ: ARMH) architecture.  Unlike Intel, ARM licenses their processor designs to anyone who’s willing to pay.  ARM states that there are over 80 licensees for the Cortex family.  This is great for consumers, but it makes manufacturing ARM processors a tough business to be in.  Furthermore, licensing an ARM processor design can make it difficult to achieve product differentiation. SoCs from different manufacturers that feature the same ARM design will generally have roughly equivalent performance as far as the core processor is concerned.
  3. The growth of SoC customization.  Manufacturers such as Apple and Samsung don’t buy off-the-shelf SoCs for their mobile devices.  Instead, they design the SoCs themselves, although that often consists of simply taking some existing component design, such as a graphics processor, and slapping it into the SoC layout.  But even this level of customization offers advantages.  The device manufacturers can decide what types of hardware interfaces they want the SoC to support and dispense with anything they don’t want.  When I look at an OMAP 5 block diagram, I see just about every hardware interface I can imagine, from USB and external memory drivers to LCD displays and WiFi.  Of course.  TI wants to appeal to the widest range of OEMs, so they throw in everything that an OEM might want, whether they want it or not.  But while this approach works fine in the PC box world, it doesn’t work as well in the mobile device world.  Power and circuit board real estate are at a premium.  Providing processing and interface capabilities that an OEM doesn’t want is often a non-starter.

The End of the Big Chip Maker? 

Much has been written about the end of the “PC Era,” and I think we may be seeing the end of the Big Chip Maker era as well.  During the PC era, the Big Chip Makers produced  their microprocessors, and the device (PC box) makers designed the devices around them.  In the Post-PC era, the order of operations has been reversed.  The mobile device manufacturers design their devices, then customize the processors to suit the device design.  In effect, processor design responsibility has been commandeered by the mobile device manufacturers, and the processor manufacturers have been reduced to a supporting “foundry” role.  

Apple (NASDAQ: AAPL) has been the leading progenitor of this approach.  Starting with the Apple A4 processor that debuted in the first iPad, Apple has claimed that their A-line processors were custom designed at Apple.  As is often the case with SoCs, such design merely entailed the selection and integration of other component designs, such as Imagination Technologies’ line of graphics processors for SoCs.  Today’s SoCs integrate many functions that, ten years ago, would have been performed in discrete components or even circuit boards.  Just as a PC manufacturer of ten years ago “designed” a PC by selecting and integrating these components, today’s SoC designer selects from a wide variety of “components” that exist merely as licensable designs until the SoC silicon is actually fabricated.  

Thus, Apple can claim that their SoCs are “custom designed at Apple” when in fact, design responsibility has become highly dispersed.  In the latest Apple A6 (and A6x) SoCs there is an indication of Apple having assumed greater design responsibility, specifically for the core processor design, but in any case, the role of the Big Chip Maker has been eliminated.  

In mobile devices, the need to conserve power and reduce chip footprint drove the role-reversal of device vs. processor.  Is it possible that this process will extend to the traditional PC?  Certainly one can contemplate that a system manufacturer such as Apple, HP, or Sony might decide that their all-in-one desktop computer just isn’t thin enough, and turn to a custom ARM SOC as a remedy.  Likewise, mobile PCs such as laptops and ultrabooks might also succumb to custom SOCs.  It is precisely this eventuality that Windows RT is intended to address.  So Microsoft has covered its rear, even if Intel has not. 

MarkHibben has a position in Apple. The Motley Fool owns shares of Apple and Intel. Motley Fool newsletter services recommend Apple, ARM Holdings, and Intel. Try any of our Foolish newsletter services free for 30 days. We Fools may not all hold the same opinions, but we all believe that considering a diverse range of insights makes us better investors. The Motley Fool has a disclosure policy.If you have questions about this post or the Fool’s blog network, click here for information.

blog comments powered by Disqus