Do the Economics of Samsung's Exynos Work?

In a prior piece, I explored the potential of Samsung Electronics (NASDAQOTH: SSNLF) dropping the development of its mobile system-on-chip product line known as Exynos. While Samsung does use the Exynos family of processors in a number of designs (notably the international variants of its Galaxy flagship phones as well as its Wi-Fi-only tablets), it's worth exploring whether the economics of this venture even make sense at this point.

Why does this matter?
A company as large, powerful, and profitable as Samsung Electronics generally isn't going to worry about the R&D costs associated with trying to develop mobile processor technologies. However, with Samsung's CFO recently offering downbeat commentary about how its second quarter went, it's likely Samsung will rethink a number of its R&D programs in a bid to keep costs in check.

One potential program Samsung could consider shutting down is its Exynos processor line. It uses its own Exynos processors in a very small subset of its vast smartphone portfolio (though tablets are a bit better), and it's tough to remember the last time a Samsung modem showed up in a device that mattered.

How much does it cost to run a mobile chip business?
It's quite tough to get a handle on how much R&D Samsung allocates to its Exynos processors given that Samsung's bread-and-butter semiconductor business is actually NAND/DRAM. Broadcom (NASDAQ: BRCM) claims that shutting down its baseband business will result in annual savings of $700 million per year.

Keep in mind, though, that it's unclear how much IP leverage Broadcom was getting from its other businesses (set top box, infrastructure) with respect to the other portions of its system-on-chip products (CPU cores, image processing, graphics, etc.). The cost of running a mobile chip business without said leverage is probably well north of $1 billion (mobile chip leader Qualcomm (QCOM 1.41%) spends north of $3 billion).

What's the best case for Exynos?
Let's assume the best case for Samsung's Exynos is that it can get these chips into all of its roughly 300 million smartphones shipped per year, as well as all of the roughly 40-50 million tablets it ships per year. Further, let's assume the following:

• In the current setup, Samsung pays an average of $20 per system-on-chip from external vendors (principally Qualcomm) across all of its products (this probably overstates how much Samsung needs to pay, but this serves as a nice upper bound for the cost savings).
• Samsung's per-die cost to manufacture its own chips at its wafer factories is $6.
• Samsung's mobile chip R&D would need to increase dramatically to keep up with what Qualcomm is doing (let's call it $2 billion/year).
• 350 million annual mobile processors needed.

From this, let's see how the financials for both could potentially look if Samsung ramped up its R&D efforts to try obviate the use of Qualcomm's chips in its devices:

Source: Author estimates.

At first glance, it looks like the savings are actually quite compelling -- on the order of $2.9 billion per year -- assuming that these numbers are mostly accurate. That said, this best-case scenario is probably unrealistic.

The more realistic case looks neutral
If we assume Samsung won't be able to transition to all-internal for its mobile devices, and that Exynos tops out at about 30% of all shipped Samsung devices (which is up from the 10%-20% that Imagination Technologies estimated on its most recent earnings call), the picture looks dramatically different.

Source: Author estimates. 

Here the assumption is that R&D can come down a bit as Samsung doesn't need to worry about developing as many SoCs (but still has to develop the foundational IP as well as a number of designs). When all is said and done, it looks like if Samsung can get about 30% of its apps processor consumption from its own teams, then It's a wash between all-external and an internal/external hybrid. 

Foolish bottom line
Keep in mind that the estimates presented here are very sensitive to the assumptions for Samsung's manufacturing costs as well as the selling prices of chips from Qualcomm (and others). That said, it does seem as though Samsung could stand to benefit if its in-house chip teams can pull ahead of what chip specialists like Qualcomm can provide. That's a big "if," though, and something that Qualcomm -- as well as other Samsung suppliers -- will work very hard to keep from happening.