Recently, I began vetting NVIDIA (NVDA) for my own personal investment. Through my own research, I've gone from bullish to bearish inside of a month.
I was led to NVIDIA by the success of Google's (GOOG) Nexus Tablet, a device that makes use of NVIDIA's Tegra mobile processor technology. And, based on this success, the stock staged a nice run from the mid 12s to the mid 14s before pulling back a bit - not a bad run in the short term. The company also finds some prized wins in Apple (AAPL) desktop and notebook products, so it looked like a real winner on the surface.
But I started inspecting their most recent quarterly report to find some troubling items when presented in the light of mobile computing trends - largely the very same problem that Intel (INTC) has with their PC processor business. To start, for the six months ended July 29, 2012, the bulk of their business - 63.4 percent - came from their desktop and notebook products (referred to in the report as the "GPU" segment). Not surprisingly, this is down from 64.5 percent year-over-year. But, hey, this represents the success of their mobile products, right? Not so - the top line in this segment is down 2.2 percent.
For me, it all came together at this week's Intel Developer Forum, where a demonstration of the upcoming Haswell integrated graphics was given. This demonstration was impressive, to say the least. While it is important to note that integrated graphics have been widely expected to eventually displace a large portion of the discrete graphics products (like NVIDIA's) for the last decade - this is why AMD (AMD) purchased ATI, after all - the realization has always been just over the horizon and seemed to be staying there.
Because new, power-intensive features were always being added to 3D computer rendering technology, it always made sense to generate discrete graphics products to satisfy the cutting edge of the market. Competitive product cycles were churning out new products every six months while the CPU industry was at 18 to 24 months. As such, it never made any sense to combine these two. Integrated graphics have been around for more than a decade and they have not impacted the sales of NVIDIA's premium parts. With Intel's demonstration this week, I think that those days are now numbered and integrated graphics are going to begin encroaching on NVIDIA's bread and butter PC and notebook products, which, at best, is a shrinking segment to begin with.
NVIDIA, having already established themselves in the mobile processor business, are definitely ahead of this realization - a new segment to replace the erosion in the GPU business. But they are competing with Qualcomm (QCOM), Texas Instruments (TXN), Samsung (SSNLF.PK), Marvell (MRVL), Intel and Apple. Should we expect margins to be anything but razor thin? The real winner here is ARM (ARMH), which licenses a power-efficient, copy-and-paste alternative to Intel x86 architecture - an architecture which is only still popular because Microsoft (MSFT) Windows PCs required it up until Windows 8. Yes, Intel has a post-PC problem, too. But this article is about NVIDIA.
NVIDIA spent the last fifteen years furthering their brute force PC graphics technology. Power and bandwidth efficiency were never important to their customers. Subsequently, their products ended up looking like small vacuum cleaners in order to prevent thermal meltdown. NVIDIA and ATI made billions with this approach, displacing a more efficient technology from the PC industry: PowerVR by Imagination Technologies (IGNMF.PK) - a company that now licenses copy-and-paste graphics technology to the myriad of mobile processor suppliers.
Buy an ARM processor license and a PowerVR graphics license, stick them together like legos in any scalable multiple and then have a fab-for-hire like TSMC (TSM), Samsung or Global Foundries produce the chip. While technically complex, there's not a lot to this - and that is why the mobile market is flooded with margin-bleeding competition. As an aside, it is important to note that Qualcomm has purchased a tile-based, PowerVR-like graphics technology from AMD after it became apparent that this was the best way to do mobile graphics. They could have struck a deal with NVIDIA if they believed it to make any sense.
With Intel entering an already very crowded mobile processor market, I believe that NVIDIA's desktop and notebook "bread" is getting stale beneath a very thin layer of mobile "butter."
Disclosure: I have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.