Two great American companies are soon to engage in some "Bump and Rub," as they say in the racing world.
Intel (INTC) is the leading and largest true and pure play semiconductor manufacturer on the planet. Intel makes integrated circuits intended for computing applications at all levels. As such, the Intel strategy is simple: Supply the finest high technology computing products to all equipment manufacturers.
Qualcomm is the world largest fabless semiconductor company. The company specializes and virtually "owns" the baseband radio section in a large percentage of the dumb, feature and smart phone market. Qualcomm is also one of the leading suppliers of ARM (ARMH) based application processors to the smart phone and tablet market. Qualcomm buys their semiconductor products primarily from TSMC (TSM), and is beginning to use other foundries such as Samsung and Global Foundries, due to TSMC's inability to supply all of the Qualcomm product requirements. The IP (Intellectual Property) division of the company is about a third of revenue but produces nearly three quarters of the operating profit.
As a late starter, Intel has been at a disadvantage in two areas. Intel does not have a 4G LTE solution today and the company has been thought to be lagging in low power consumption of their x86 application processors. The 4G LTE solution is expected to be here by the end of the year. I would forgive Intel if the LTE debut was delayed until CES in early January. The power issue was never a real issue, since the single core Medfield was competitive with the best in class in real world applications. Intel is expected to announce a dual core Medfield SoC along with the LTE chip. I would expect the dual core Medfield will be tweaked for power so that it beats all competition on power consumption. The mule to carry these parts to a worldwide market is likely to be a Motorola/Google (GOOG) 4G LTE smart phone with features superior to the Apple (AAPL) iPhone and the best from Samsung. A perfect venue to announce such a smart phone would be the upcoming CES.
With Intel shipping a low power AP and 4G LTE solution, Qualcomm will be competing directly with the world's leading semiconductor producer with no functional or cost advantage.
Here's the problem for Qualcomm: Qualcomm generates 64% of revenue and about 25% of operating income from the sale of application processors and radio chips. The company generates 32% of revenue and 71% of operating income from patent licenses and royalty payments. Qualcomm doesn't need the semiconductor division to be a very successful company.
To make matters worse for Qualcomm, the Intel parts will be made on fully depreciated 32nm production lines while the Qualcomm parts will be coming from a low yield, high cost TSMC fab. The Intel parts could sell to the end customers for a price that is below the Qualcomm cost from TSMC. To make matters even worse, if that is possible, Qualcomm will be able to see and know what is in store for them when Intel moves to a new Atom architecture SoC built on the 22nm Trigate process later in 2013.
Qualcomm will soldier on in order to see if TSMC can actually build anything on a planar 20nm process. This should be known by June of 2013. The chance of TSMC success is low due to the leakage problems with a planar process at 20nm (that's why Intel moved to Trigate). TSMC talks about running a 16nm "finfet" process, but not until 2015. Even if successful, by that time, Intel will be ramping their 10nm node, so TSMC/Qualcomm will still be 1+ node behind and, therefore, significantly behind on speed performance and power performance, never to catch up.
So, after six months of "Rubing and Bumping," if TSMC is unable to deliver 20nm sometime in 2013, Qualcomm has a giant decision to make. That decision is whether to even stay in the semiconductor business or pull a Texas Instruments (TXN) and withdraw from the mobile business altogether. TI stopped leading edge process development at 65nm, deciding that competing with Intel just created financial nightmares. TI further decided that they, in combination with TSMC, would still be at an ultimate disadvantage with regard to Intel. Qualcomm could well make the decision to withdraw, since they would still retain a sweet IP business, and by the time the "Moment of Truth" arrived, Intel could have priced Qualcomm into a loss position on their fabless semiconductor operation.
The reverberations of such a decision would be violent and wide ranging. TSMC would instantly lose their largest fab customer. TSMC would then be faced with the same decision that Qualcomm was faced with.
Apple, if they had not already made the decision to move their fab business to Intel, would now be crystal clear on that issue.
With Qualcomm and TSMC gone, Apple at Intel, the Samsung Semiconductor division would find itself in big trouble. The Samsung smart phone division could well decide to use Intel for the brains in their products.
Within a year, Intel could go from "Zero to Hero" in the mobile semiconductor business.
Maybe this is what Morris Chang was thinking when he thought losing the process technology race to Intel was "horrible" to contemplate.
The move here is to buy Intel and sell Qualcomm. Also, owning the overpriced ARM Holdings would not lead to rocking chair money.