On Tuesday, Texas Instruments Inc. (TXN) announced that it would "reduce R&D expenditures" on the OMAP SoC used in many smartphones and tablets, including the latest Amazon Kindle. Decoded, Texas Instruments just told the industry it was exiting the mobile application processor business.
By virtue of the announcement, we can be assured that Texas Instruments will never get another design win for a mobile OMAP chip. Why would Texas Instruments leave the mobile business, while Intel is spending huge sums of money to enter that same business?
Years ago, Texas Instruments decided to leave the leading-edge manufacturing race and become primarily an analog products supplier. Analog doesn't need anything near a 22nm process, as large chip processors do.
So, in the OMAP product for the mobile business, Texas Instruments had become a fabless supplier with Taiwan Semiconductor Manufacturing Company Ltd. (TSM) producing the OMAP Wafers for them.
Texas Instruments' latest move has no significant immediate affect on its competitors. Over time, the Texas Instruments business will simply be absorbed by the remaining players.
Who are the remaining players and, more important, who wins and loses from here?
Samsung is in the enviable position of manufacturing the Apple (AAPL) A6 chip, as well as its own smartphone chip requirements and some third-tier business. Since Samsung smartphones will outsell Apple about 2:1, and some projections for the Apple iPhone volume for the next year are in the 100 million unit area, Samsung could be the manufacturer for 300 million mobile SoC application processor chips. That is approximately 50% of the market.
Qualcomm is the second place player due largely to its position in the baseband radio modem business and particularly its leadership position in the LTE modem. Nvidia is the third place player fighting hard to catch up with Qualcomm. Its LTE chip will be available later this year. Qualcomm and Nvidia probably share 30-35% of the application processor business (remember, Texas Instruments still has 15-20% of the current business, going to zero in the next two years).
Bringing up the distant rear is Intel. Intel is dabbling in the smartphone business with offshore customers. Intel still has no position in smartphones for the US market because it doesn't have LTE capability. That all changes in the very near future. Intel claims it will be shipping a dual core Atom-based SOC application processor by the end of the year. It also claims that an Intel LTE chip will be available "very soon". Some speculate the Intel LTE function will be integrated on the new dual core chip.
From a manufacturing technology standpoint, Samsung is self-contained, but about four years behind the leader Intel. Qualcomm and Nvidia are fabless and utterly dependent on Taiwan Semiconductor Manufacturing Company for their wafer manufacturing and Taiwan Semiconductor has both capacity and yield issues that have caused Qualcomm and Nvidia well-publicized availability problems that have reduced sales projections.
Taiwan Semiconductor Manufacturing Company is also about four years behind Intel on leading-edge fabrication capability. If Texas Instruments, who has a good early foothold in the mobile business, feels it cannot win as a fabless suppler, what makes Nvidia and Qualcomm think they could be the endpoint winners?
Intel is the undisputed world leader in the highest volume finest geometry manufacturing processes. Depending on the source, Intel has anywhere from a four-year to infinite-year lead in manufacturing technology over the next best competitor. (Infinite in that Taiwan Semiconductor and Samsung might find it uneconomical to try to keep the node competition going, a la Texas Instruments).
With a low power dual core atom SoC and either discrete or integrated LTE capability available yet this year, Intel could pick up market share against the other three in quick-time.
By the way, can anyone really envision Apple still using Samsung for its A6 or A7 chips a year from now? I am amazed at how quickly Apple has virtually eliminated Samsung as a display supplier, battery supplier, and memory supplier. The "A" chip has to be next to go.
Intel is then the only other manufacturer with the capacity and the technology to produce Apple's A chips whether they are ARMH-based or Atom-based.
Given the above, Intel has a good chance to move to first place in the mobile space in the next year or two. The mobile application processor market (including LTE) should be about $14 billion by that time, excluding the captive Samsung business. Intel could get $7 billion of that. My opinion: buy Intel, sell all the others.
Disclosure: I am long INTC.