According to a new report from DigiTimes, Samsung (NASDAQOTH:SSNLF) will "dump Qualcomm's (NASDAQ:QCOM) Snapdragon 810 processor for use in its next generation Galaxy S-series flagship smartphone, in favor of its own Exynos-series chip, due to concerns related to the Snapdragon 810 overheating issues."
The report further asserts that Samsung will, at first, use its internally designed 14-nanometer-based Exynos processor, but that it will "gradually increase the ratio for devices using Qualcomm's processor."
This may actually be the strangest rumor that I've heard about the Galaxy S6 yet.
Why would Samsung move backward?
The Snapdragon 810 processor is known to be built on Taiwan Semiconductor's (NYSE:TSM) 20-nanometer manufacturing technology. The Exynos 7420 that Samsung will reportedly be using in the Exynos-powered variant of the Galaxy S6 is built on Samsung's more advanced 14-nanometer manufacturing technology, which should provide meaningful efficiency and performance benefits over the 20-nanometer TSMC process.
Additionally, Samsung recently made a relatively big deal about its category 9-capable Galaxy Note 4, which reportedly features a Samsung-designed cellular solution and an Exynos processor. If this is the case, then according to the picture painted by Samsung, there is no real need for Qualcomm's processors since it has access to more advanced manufacturing technology and seemingly just-as-good baseband technology.
And yet, with all of this in mind, Samsung is still reportedly looking to shift mix more toward Qualcomm chips as launch schedules permit.
Why would Samsung want Qualcomm chips built at TSMC?
As strange as it seems for Samsung to want to move to Qualcomm chips built on an inferior technology node, it's even stranger that Samsung would shift to chips built on Taiwan Semiconductor's process. Every integrated Snapdragon chip that Samsung uses is a leading-edge, Samsung-built processor and possibly modem that isn't used.
Given that Samsung is apparently taking its foundry business quite seriously, it seems strange to shoot its own foundry business in the foot while at the same time giving its fiercest competitor volumes to fill its factories.
That said, there is a potential explanation that works. Samsung, per IDC, had about 23.8% smartphone market share in the third quarter of 2013 -- a number that's declined from last year's 32.5%.
If we assume also that Samsung's Exynos chips will see limited traction outside of Samsung smartphones, then that effectively caps Samsung's long-term chip opportunity to Samsung phones. Samsung's foundry division would essentially depend on the execution of Samsung's mobile device division.
However, if Samsung keeps the door open to Qualcomm -- which, per Strategy Analytics, owns over 50% of smartphone applications processor revenue share -- and allows it to eventually join the Galaxy S6 supply chain once the alleged issues with the Snapdragon 810 are worked out, then Qualcomm might be more comfortable transitioning more of its chip volumes to Samsung over time (Qualcomm has built cellular baseband processors at Samsung in the past, per Chipworks; I'm not sure if it has ever built a Snapdragon processor).
In fact, according to a report published today, Qualcomm has reportedly stopped trial production of 16-nanometer parts at TSMC, with DigiTimes claiming that yield improvements on Samsung's 14-nanometer are the reason.
Samsung may believe that keeping the entire smartphone chip market open from a foundry perspective to it at the expense of its Exynos chips is the right way to go from a long-term perspective.
Let's wait and see
I will be keeping a very close watch on any developments in this space. The Galaxy S6 launch likely isn't too far away now, and the tear-downs that will invariably appear after the device's launch should shed some more light on what exactly is going on with Samsung, Qualcomm, and the Galaxy S6.