Nvidia (NVDA), known in the past mainly as a maker of graphics chips (GPUs) used primarily for PC gaming, is struggling with the market transition set in motion by consumer demand for smartphones and tablet computers. Nvidia generates significant amounts of cash each quarter, has a large cash reserve, and has an enviable cache of IP (intellectual property). Despite that, revenues have not been growing, the stock price is stagnant, and competition has intensified.
The slump in the traditional PC market (servers, desktops, and notebooks) is well-documented. Nvidia has not suffered as badly in the PC market as Advanced MicroDevices (AMD) and Intel (INTC) because its GPUs tend to go into gaming and production-oriented machines, which have not been as hurt by the switch to tablets and smartphones for consumption-oriented computing. As 86x CPU and GPU integration accelerates, however, even most production-oriented machines may work well without the added cost of an extra graphics card. It is at least possible that younger people will come to prefer gaming on small screens and not feel they need to invest heavily or regularly in large-screen gaming.
Nvidia's response to the transition has been to take its graphics IP and add it to a CPU with an ARM architecture, under the brand name Tegra, targeted at smartphones and tablets. This worked reasonably well for Tegra 3, but sales have fallen off while waiting for Tegra 4. One version of Tegra 4 should be ready for smartphones by incorporating a cellular transceiver (which previously had to be on a separate chip). This will help, but other companies already have working SoCs (System on Chip) incorporating a CPU, GPU, and transceiver, so device makers have other choices.
I just looked at an ad for the new version of Google Nexus 7 tablet, and it boasts use of a "a quad-core Qualcomm Snapdragon™ S4 Pro processor." The first generation Nexus 7 used an Nvidia Tegra 3 processor, also quad-core. That is an important slot to lose, one Nvidia was boasting about last year. I have no doubt there will be announcements of tablets and smartphones based on Tegra 4, but I do think the competition is intense. The basic ARM designs can be licensed by any semiconductor chip maker. Nvidia's graphics advantage (and AMD's as well) apparently is not as important on small screens as on big screens. Over the last decade a lot of engineers have learned the secrets of graphics processing, so building or licensing a graphics unit for a smartphone or tablet SoC is not as big of a deal as it was even five years ago.
How much cash can various chip makers throw into the competition? Samsung's resources are presumably up there in the stratosphere with Intel, with the additional advantage of being able to feed its own SoCs into its own-branded tablets and smartphones. Here's how some of the top U.S. competition shapes up:
|lastest reported Q, |
|Cash & equivs. balance||cash flow from operations||R&D expense|
Each company has a different set of products addressing different end markets. Intel's money comes from x86 CPUs. Intel has struggled to enter the tablet and smartphone market. AMD's lack of money is from x86 CPUs and GPUs, is not even trying for the smartphone market and has struggled to get any tablet slots. Qualcomm's (QCOM) cash come from cell phones and smartphones, where it is the chipmaker to beat or from which to take market share. Marvell's (MRVL) cash is mainly from HDD controllers. Marvell has a substantial share of the Chinese smartphone market and has done well with set-top boxes, but has not done as well winning significant tablet slots.
Clearly Intel has the cash to stay in the race for the mobile market as long as it wants to be. AMD is likely not a contender at the moment. Qualcomm has the market share, cash balance, cash flow, and R&D budget to dominate the space. Qualcomm is not likely very worried about anyone but Samsung. Even Marvell is reasonably well-equipped to compete with Nvidia, at least in niches. Apple (APPL) is essentially in its own category for now, since it does not make chips or software for other device makers. For the smaller players like Nvidia to increase R&D to Qualcomm-like levels would involve joining AMD in negative cash-flow territory.
The problem for Nvidia is obvious. Nvidia lost the game console business, the independent GPU business will likely dry up over time except for very high end systems, and competing in the smartphone or even tablet space is not easy. It is not like the old days when the main worry was ATI (now a division of AMD) or vague threats from Intel.
A lot is riding on Tegra, but Tegra has no clear long-term advantages over other rivals. Tegra revenue declined rather dramatically, 71% y/y, for the July quarter. This is partly to transition to Tegra 4, but going back to 2011 or 2012 Tegra was being touted as the clear path out of the decline GPU market dilemma. In August 2012 management boasted that Tegra had a spot in the coming Nexus 7 and Microsoft (MSFT) Surface RT. Now Qualcomm has the Nexus slot and Surface had minimal sales. I am certainly watching for slot win announcements for Tegra 4, but I won't assume slot wins translate to sell-through. Also keep in mind that there are now many smaller smartphone chip makers in the competition mix, notably in China.
I'm not suggesting Nvidia abandon Tegra; enough has been invested to go ahead and see how it works out, but looking down the road 2 or 3 years, I don't see smartphone graphics processing as being anything but a commodity. There will be niche areas too, but they will be fought over as well.
In the long run, Nvidia's stockholders might be better served by Nvidia becoming a niche player focused on supercomputers and datacenters, as well as graphics cards for workstations. I am beginning to think that adding Nvidia's high-end graphics designs to the coming ARM-based server designs will work out better than pursuing the smartphone market. It means going back to AMD and Intel as primary competitors, but Nvidia has always done well competing with AMD and Intel.
On August 8, 2013 Nvidia reported on the quarter ending July 28. Revenues were $977.2 million, up 2% sequentially from $954.7 million, but down 6% from $1,044.3 million in the year-earlier quarter. GAAP net income was $96.4 million, up 24% from $77.9 million but down 19% from $119.0 year-earlier. GAAP EPS (earnings per share) were $0.16, up 23% sequentially from $0.13, but down 16% from $0.19 year-earlier.
It is notable that the Quadro professional graphics segment showed "strong" sequential and y/y growth. It shows the PC is still important to those who use it for computation-intensive or graphics-intensive work. This quarter the GRID GPU chips for graphics for servers (including cloud servers) are going into production at Cisco (CSCO), Dell (DELL), HP (HPQ), and IBM (IBM). In addition, Nvidia plans to start licensing its IP more (it already licenses to Intel), which should improve margins, at whatever time that actually happens.
Given the August 19 closing price of $14.95, I would call NVDA attractive short-term (1 to 4 quarters), but with long-term dangers. Seasonality and the launch of GRID and Tegra 4 should give the Nvidia a nice revenue and profit bump in the October and January quarters. Trailing 12 month P/E is 16.9, reasonable in this market. There is certainly a possibility I am overly gloomy about Nvidia's prospects with Tegra.
But since guidance is for October revenue near $1.05 billion, management is not predicting a big Tegra 4 bump. Note also that it represents a 12.5% decline from $1.2 billion in the October quarter of 2012.
Nvidia would become very attractive, short term, at a price under $15, if Tegra revenue ramps enough in the April quarter and beyond to overcome normal seasonality. Since Nvidia plays a dividend of $0.075 per quarter, about 2% a year, I think the down-side risks during the next year are minimal.
By 2015, just about everyone who is within range of a cell phone tower will have a smartphone. It is a huge global market. I just don't think there are going to need to be that many players around when the industry consolidates.