I haven't written about Nvidia (NVDA) in a while, primarily because I was waiting on the Intel (INTC) quarterly results (as well as some other data-points) in order to try to get a glimpse into how the company's core GPU business is doing. Following the Intel report and some observations in the marketplace, it seems clear to me that Nvidia should continue to perform quite well as we get into the back half of the year. I expect dGPU share gains, particularly in the notebook space, the ramp of Tegra 4 in a number of key designs, as well as continuing strength in the professional/HPC GPU market. I also remain positive on the company's capital structure, and in particular, the buyback program.
The Discrete GPU Space: Minimally Affected By Weak PC Sales
A point that I typically try to emphasize in my Nvidia pieces is that while Nvidia's GPUs are primarily sold into the PC market, the company is shielded (no pun intended) from much of the decline as it plays primarily in the high end space. If you listen to the most recent Intel call, you will notice the following important snippet,
And if you look at the last three years even in a time where the market's been relatively weak, the Core i5, i7 volumes have been very healthy and if anything we've seen a larger mix overall to Core over that time period.
This suggests that the brunt of the PC decline is actually happening in the low end Pentium/Celeron segment of the market, which is precisely the market that Nvidia's discrete GPUs do not play in (b." ut Nvidia's Tegra helps to eat into via tablets). The major headwind that I see here is that integrated GPUs on the Intel side continue to improve, but the argument (and I believe this wholeheartedly) is that as games continue to evolve and become more realistic, there will not - for the foreseeable future - be a "good enough". When a 2W mobile SoC can render photorealistic, infinitely-spanning worlds at 60 frames per second, then it will be time to panic about the integrated GPU threat.
Also, as I have tracked the major new gaming notebooks that have hit the shelves, particularly the Razer Blade, MSI GE40 and other "Haswell" based gaming-oriented machines, I'm seeing a rather substantial number of design wins for Nvidia, and limited adoption of AMD's (AMD) Radeon or Intel's own "Iris Pro" integrated GPU. I suspect the former is due to the fact that Nvidia's parts are generally more power efficient as well as superior on the software side of things (Optimus works much better than Enduro, for instance). The latter is likely explained by the fact that despite Intel's GT3 graphics exhibiting a significant amount of compute horsepower, it often falls short in actual gaming due to both the design of the GPU (compute heavy; graphics specific structures not as prominent) as well as the drivers (Intel's have become much better, but still are no match for Nvidia's.
The only thing that gives me a headache is that Apple (AAPL) is likely to switch away from Nvidia's discrete GPU in the next iteration of the 15" Retina MacBook Pro and will likely utilize Intel's Iris Pro. This trade-off is not likely to be particularly widespread as Windows PC buyers paying for a discrete GPU very likely care about gaming performance more than form factor, but such is not the case on the Apple side of things. I do, however, expect that this is well known and baked into the share price.
On the desktop side of things, I refer you once again to Intel's report and note the following tidbit,
- Desktop platform volumes decreased 3% from Q2 2012 to Q2 2013
- Desktop platform average selling prices increased 6% from Q2 2012 to Q2 2013
While the death of the desktop PC (where many of Nvidia's GPU margin dollars come from) has been a hot topic for debate, it seems that the desktop space has essentially bottomed, and the mix is richer. This suggests that, once again, the higher end of the market is performing quite well (and is likely to get a boost with the release of the "Ivy Bridge E" platform), which bodes well for Nvidia's higher end (and high margin dollar) GPUs. The message is clear: while the PC end market as a whole is soft, the high end is doing okay, and this means that Nvidia's GPUs - which are exclusively higher end oriented - are fairly protected from the secular decline in the low end.
Tegra 4: Finally Seeing Some Design Win Traction, But Tegra 4i Is the Real Story
It is encouraging that there are some Tegra 4 designs shipping today, particularly from Toshiba with its Excite Pro and Excite Write lines. I am also encouraged to see that HP (HPQ) chose Nvidia's Tegra 4 for its SlateBook X2 convertible Android design, and it is now coming in that ZTE is planning on using the Tegra 4 in its upcoming "Geek" phone. I expect further tablet traction as we go through the year, and I would not be surprised if Microsoft (MSFT) continued its partnership with Nvidia for the next generation of Surface RT (although I admit to having modest expectations for sales on this as Bay Trail tablets are likely to prove to be the better value proposition for most consumers).
While Nvidia claims that it was late with the Tegra 4, it is clear that the chips are showing up in designs in time for the back-to-school season and will, more importantly, be available during the prime holiday selling season for tablets. I expect further design win momentum as we go forward, particularly on the Android side of things (I do not expect any third party vendors to build Windows RT devices), and I would not be surprised even to see Tegra 4 powered Chromebooks in the very near future.
However, despite the fact that Tegra 4 has a pulse, I am much more excited by the upcoming Tegra 4i part which is a quad ARM (ARMH) Cortex A9 r4 with a 60 shader Nvidia-designed GPU and an integrated Icera (owned by Nvidia) LTE modem. This part could be positioned quite well to capture share in the high volume smartphone segments, particularly as management has indicated that this will be a very small chip that can likely be sold profitably for cheaply. The traction for this part should be clear by Mobile World Congress 2014, and from a long-term perspective remain positive on the asymmetric risk/reward opportunity that Nvidia's entry into this space represents.
Professional GPU And HPC
The professional GPU business seemed to bottom a couple of quarters ago as the Kepler refresh, coupled with the Romley ramp, drove a recovery. I expect continued growth in this space, although I find it frustrating that Nvidia will no longer be reporting the professional GPU segment as a separate item in the financial reports. I remain optimistic, however, that Nvidia is taking full advantage of the rebound of this sector, and I further believe that it is unlikely that competitor, AMD, took any meaningful share in the workstation graphics.
In HPC, Nvidia is still likely to continue to see nice growth, but for the first time it has a very serious competitor with Intel's Xeon Phi. While I do not believe that the first generation Xeon Phi is sufficiently competitive with the Kepler architecture in this space to drive meaningful share loss (although there have been a number of headline "wins" for the Phi parts), I believe that going forward Intel will be a very competitive player in this space, starting with Intel's "Knights Landing" part which promises to triple performance per watt and even introduce standalone/socketed designs. This, fortunately for Nvidia, is likely to be in the late 2014/early 2015 timeframe according to leaked roadmaps, so I expect Nvidia to be competing with its Maxwell part, which will be substantially improved from the current market-leading Kepler.
I remain positive on Nvidia, and I believe that the company's best days are still ahead of it. With an improving competitive position in mobile, a stronghold in the traditional GPU space, and a dirt cheap 5.36x EV/EBITDA valuation, I believe shares are still worth buying on dips and remain long on the name.