Take a look at the mobile system-on-a-chip, or SOC, landscape today -- in particular on smartphones. What do you see? You see Qualcomm (QCOM +8.17%), essentially owning the high-end market; Qualcomm and MediaTek in a tug-of-war for high-volume mass-market phones in Asia, and a number of other Asian vendors such as Allwinner, Rockchip, and Spreadtrum fighting for the scraps. It's a brutal business.
Why is it so brutal? Barrier to entry. For the low-cost mainstream devices, these chips need to be cheap; and within a fixed price, they still need to offer as much functionality and performance as possible. Thanks to the ubiquity and quality of off-the-shelf intellectual property, or IP, from ARM Holdings and Imagination Technologies, the focus is on enhancing the non-computing parts of the SOC for both functionality and low cost (think video decoding blocks, modems, connectivity, and the like).
In this business, if you over-design your chip and end up with one that's too expensive (or, worse, late), then you're not getting the sale. If you under-design the chip and end up lacking on performance or features, you, once again, don't get the sale. With so many competitors duking it out with much of the same IP, getting your SOC wrong is deadly to your business.
The market share numbers have spoken
According to Strategy Analytics, here were the top apps processor vendors by market share in 2013:
|
Vendor |
Smartphone Share |
|---|---|
|
Qualcomm |
54% |
|
Apple |
16% |
|
MediaTek |
10% |
|
Others |
20% |
In the "Others" category you have the following vendors (this list is not exhaustive):
- Broadcom
- Intel
- NVIDIA (NVDA +1.73%)
- Spreadtrum
- Allwinner
- Rockchip
- ST-Ericsson
The overall market is still growing, and, frankly, a very strong and focused player could wrestle away share from Qualcomm, or, to a lesser extent, MediaTek, but it's not for the weak-hearted or those without the willingness to eat gross/operating margin dirt in order to make real headway.
NVIDIA no longer focused on mainstream phones
With the Tegra 4i (codenamed "Grey"), NVIDIA had hoped to make a real dent in the value smartphone market. With an integrated LTE baseband (courtesy of the company's Icera acquisition), and a pretty solidly performing part, there was a lot of hope that NVIDIA could become a real competitor to Qualcomm in lower-cost phones. It seems that NVIDIA believes that this market is no longer one it should be spending its resources fighting for.
Indeed, the following question and subsequent answer from NVIDIA CEO Jen-Hsun Huang essentially confirms that the Tegra 4i could perhaps be the first -- and last -- NVIDIA effort for mainstream phones:
Analyst: So, a lot of detail on virtually every segment of your business, but not too much on the cellular front. I was wondering, is that an integral part of the Tegra strategy and do you need cellular to continue in your roadmap?
Jen-Hsun Huang: So, we also didn't talk about PC OEMs today. And let me tell you why, because in the final analysis those are design wins; they come and go. Let me give you some -- I'll say some things I believe. Now, we are going to get design wins in tablets in very important companies, super-phones in important companies, and we're going to continue to. I believe that the mainstream smartphone market is saturated and it's not worthwhile to pursue.
It doesn't get much clearer than this, and, frankly, NVIDIA is likely better for it, as it doesn't really need to win in the mainstream smartphone market for Tegra to be successful. Remember: NVIDIA's R&D investment in Tegra is about $600 million (per GTC 2013), and so at 50% gross margins, the company needs $1.2 billion in sales to break even, and then from there the business is at scale.
Can tablets, automotive, and other markets help Tegra win?
So the question is whether Tegra can hit that $1.2 billion in revenue mark to break even and from there grow into profitability. The tablet market is, according to IDC, going to come in at about 260 million units, and assuming NVIDIA can capture 10% of that market at $25-per-unit average selling price (remember: NVIDIA is playing only in the high end), that gets us halfway there.
The other half is going to have to come from automotive and perhaps other applications. It's anyone's guess as to whether NVIDIA can grow its non-tablet Tegra business to north of $600 million a year, but management likely wouldn't continue to invest in Tegra if it didn't believe that it could eventually get the business to scale.
