For its latest quarter, ended January 26, Nvidia (NVDA) reported flat revenue growth (3% y/y) at $1.144 billion. While the GPU side of the business was up a healthy 14% y/y to $0.947 billion in revenue, the Tegra Processor business collapsed by 37% to $0.131 billion in revenue. Nvidia had invested heavily in Tegra, and this was reflected in a 9% operating income decline to $166.9 million.
A Growth Stock?
Despite the terrible earnings news, its clear there are many investors who believe that Nvidia is poised for future growth. The stock popped almost 4% after hours. There's some basis for this in Nvidia's Tegra K1. Equipped with the most impressive graphics and gaming capability of any mobile processor, as well as a custom designed 64 bit ARM core, the K1 looks to be the go to chip for Android tablet makers looking to match or beat the Apple A7, as well as keep up with Intel's Bay Trail. Nevertheless, I remain pessimistic about Nvidia's long-term prospects.
In May 2013 my article "Nvidia Stumbles on the Next Generation Tegra" was greeted with some incredulity. Once again, I was challenging conventional wisdom, and it didn't go over well with Nvidia supporters who accused me of harboring some, to put it mildly, antipathy towards Nvidia. I like Nvidia graphics cards a lot and have a half dozen of them scattered about the various PCs in my household. My views on Nvidia as an investment have nothing to do with any personal feelings, pro or con.
In retrospect, it's clear that Nvidia did stumble with the Tegra, delaying the roll out of the Tegra 4 well into 2013. At the end of Q1 2013, Nvidia had stated that delaying the Tegra 4 was intentional in order to accelerate production of the Tegra 4i, which contains a 4G modem and is targeted for high end smartphones. Almost a year later, the first products with the 4i are only expected to be announced this quarter.
Only two major devices adopted the Tegra 4, the Microsoft (MSFT) Surface 2 and HP (HPQ) SlateBook X2. It's not clear how the X2 is doing, but the Surface 2 seems to have bombed in calendar Q4, with an estimated 1.275 million units sold (including Surface Pro) and Microsoft posting a loss of $39 million on Surface sales.
Countering the Intel Threat
Although the threat of Bay Trail to ARM manufacturers like Nvidia failed to materialize in 2013, it clear that Intel represents a looming threat to ARM processor makers like Nvidia. Intel (INTC) is going after Android and Windows device makers this year with subsidized Bay Trail and post Bay Trail mobile processors. Intel has announced these "contra revenue" subsidies designed to bring the mobile device chipset cost online with ARM based counterparts.
Nvidia's response appears to be to move up market with the K1. The K1 contains 192 Kepler GPU cores, which totally dwarf the four little CPU cores on the chip. This move by Nvidia is very reminiscent of AMD's building into their CPUs graphics processors taken from their graphics card line. One upping Intel in on-board graphics performance seems to be the way to go if a manufacturer can't match Intel's 22 nm FinFET process, which none can for the time being.
It remains to be established what the power requirements of K1 will be. It looks like a very large, very power hungry chip for a supposedly mobile processor. Nvidia has promised energy efficiency, but hasn't produced any numbers.
Nvidia has stated that product announcements for the K1 will come in the second half of the year, and I wouldn't be surprised if K1 finds use in an ARM-based gaming console as well as a high end tablet.
Paradigm Shift Victim
I see Nvidia's problems on the ARM processor front as being much more fundamental than just the competition from Intel. The paradigm shift that I describe in "The Apple/Arm Disruption of Intel . . ." also applies to stand-alone ARM processor makers like Nvidia. Increasingly, the ARM processor market has come to be dominated by integrated device makers like Apple (AAPL) and Samsung who use their own custom-designed systems on chip (SOC). This leaves makers such as Nvidia with only third-tier customers such as Microsoft and HP (in terms of market share).
In the long run, I see the stand-alone processor makers only surviving as part of the keiretsu of a device maker. Nvidia has yet to align itself with any particular device maker, and doesn't appear to want to. Nvidia's Shield does raise the prospect of Nvidia trying to become a device maker itself, but how many Shields has Nvidia sold? I don't know either, but it did use the Tegra 4.
Graphics Market Fluke
Nvidia surprised me with the strength of its graphics segment performance, but this was something of a fluke driven by the new found popularity of AMD's R9 290 series graphics cards for Litecoin (a crypto-currency similar to Bitcoin) mining. According to a review by Anandtech, the R9 290x achieved comparable performance to the top end Nvidia GTX 780 (at the time of the review), and with a list price of $550, the 290x should have been a worthy competitor to Nvidia's best. But with Litecoin miners buying up every AMD 290 series card they could find, there was no competition, and Nvidia had the high end gaming market to itself.
Also, most Haswell chips turned out to have rather weak on board graphics, creating an opportunity for nVidia to sell lots of GPUs to go into Haswell laptops and desktops. But with every new Intel processor generation, the on-board graphics get a little better.
AMD's success integrating their graphics processors with their CPUs to produce the SOCs used in the Sony PS4 and Microsoft Xbox One consoles points to the long-term future of the graphics processor. The future for the graphics processor is to be a patch of real estate on an SOC. If you're a graphics provider and aren't the one designing the SOC, you're providing graphics IP to the one who is.
Can Nvidia Survive the Future?
Right now, Nvidia is trying to become the one designing the SOC as well as providing its own graphics IP, while it milks whatever is left of the discrete graphics chip market. Of course, I don't see the discrete graphics market going away completely, just steadily shrinking over time. Becoming an SOC designer could work, but it really depends on Nvidia finding device partners for the K1 and 4i that can actually move substantial unit volume through retail channels. That doesn't appear to be Microsoft, or HP.
Failing to find the right mobile device partners for its SOCs will be devastating to Nvidia, but even the near-term promises little relief from the profit squeeze in the Tegra segment. Bringing the K1 into production will be expensive, with not much in the way of sales to offset these costs, except for the Tegra 4 and the 4i, for the first half of the year. I look for continuing declines in Tegra segment revenue through the first half.
Nvidia investors are living on the promise of ARM processor fueled growth, led by the K1. The K1 could be huge, if Nvidia finds the right devices for it. Investors will need to evaluate the product announcements Nvidia has promised for the K1 and 4i in H2 for revenue and unit shipment growth.