NVIDIA (NVDA -4.35%) had a great quarter and issued pretty stupendous guidance. There was very little, if anything, for critics to latch on to. GPU sales are doing well across the board, from gaming to workstations. Tegra has bottomed and is climbing back up with gusto, and GRID seems to have some real promise. Just how much upside it can drive is still pretty unclear, but it's all additive, so it's a win. NVIDIA's business is doing fantastically, despite AMD's protests that it's about to claw back significant share.
NVIDIA doesn't do low end
A hot trend in the mobile system-on-chip market has been toward the integration of just about everything possible, including the cellular modem and connectivity. Qualcomm (QCOM -0.48%) and Intel (INTC 2.32%) -- at least in its lower-end and mainstream parts -- seem dead-set on pursuing tight integration of communications, connectivity, and so on. This is important for the high-volume, low-cost part of the market.
However, despite the fact that NVIDIA does have an integrated part for this market, known as the Tegra 4i, it's clear that the company is much more interested in pursuing Tegra as a higher-performance chip targeted at tablets, all-in-one PCs, point-of-sale terminals, in-vehicle infotainment, and so on. NVIDIA has been quite clear that it wants to pursue sockets where graphics makes a difference. This suggests that NVIDIA would like to target the upper half of the phone market, as far as price points are concerned.
The integration question
Particularly interesting on the call was the following response to a question from the Topeka Capital Markets analyst regarding integration of the cellular modem in high-end parts:
Well, the super-phone market is really moving fast. And so integration is sometimes a good thing, and integrations sometimes hurt you. It's just a higher risk to integrate four things that are -- that need to be state of the art at the same time. And so some companies use things like Tick-Tock to articulate the benefit of strategies and methodologies to reduce your risks. Some companies like us, we call them ping-pong and we try to reduce our risk by moving one piece at a time, so then you can innovative on rhythmic and continuous basis even though the market is moving dramatically. And so integration sometimes helps you, sometimes it doesn't help you. And it's just hard to say.
Qualcomm has announced a 20-nanometer LTE-Advanced modem but has yet to announce a 20-nanometer applications processor, which would presumably feature an integrated modem. Intel, too, has signaled that even its first 14-nanometer high-end product, Broxton -- scheduled for mid-2015 -- would not feature an integrated modem. Only at the very end of 2015 would it have an integrated 14-nanometer part for the low-end and mainstream of the market.
Further, it's interesting to note that Apple, which is the king of high-end smartphones, continues to very heavily invest in beefing up its applications processor but doesn't seem to be serious about integrating connectivity or communications. If it were, it very likely would have purchased the Renesas Mobile asset, which included a very nice, carrier-certified LTE modem. Perhaps NVIDIA and Intel are on the right track by not integrating modems in their highest-end products?
Foolish bottom line
NVIDIA's core competency at the end of the day is building world-class GPUs. It would stand to reason that the best way for the company to carve out a profitable niche for itself in the very crowded mobile apps processor market would be to focus on having the world's best graphics in a mobile chip. This doesn't lead NVIDIA to having huge amounts of market share in mobile, but it will be able to attack higher-margin/higher-ASP products and have a more defensible position. NVIDIA's size is such that winning this high-margin business would be more than enough to drive the top and bottom lines significantly.