- Nvidia's recent shift away from mobile for its Tegra ARM processors amounts to an admission of failure.
- Nvidia had been fighting the trend in ARM processors toward customization by device makers such as Apple and Samsung.
- 64 bit processors from Apple and Intel dashed Nvidia's hopes of positioning Tegra as a premium mobile processor.
- The financial consequences of the Tegra failure may not emerge until Q4 of this year.
In his recent CNET interview, Nvidia (NASDAQ:NVDA) CEO Jen-Hsun Huang pointed to the rapid "commoditization" of smartphones for the reason Nvidia is now shifting its focus away from mobile devices to automotive and gaming platforms for its Tegra ARM processor line. In fact, Nvidia's failure in mobile was more the direct consequence of the arrival of higher performance 64 bit mobile processors from Apple and Intel. Intended as a processor line for premium segment smartphones and tablets, Apple (NASDAQ:AAPL) and Intel (NASDAQ:INTC) simply took that market away from Tegra.
Warning Signs Unheeded
The signs that Tegra was having a difficult time in the marketplace have been there for some time. In May of last year, I wrote that Nvidia had stumbled badly by delaying the release of the Tegra 4, something that Huang reluctantly admitted in the CNET interview. In "Nvidia's Tegra Business Collapses" I pointed to the 37% y/y decline in Tegra revenue (for the fiscal quarter ending January 26, 2014) as a sign that Tegra was in serious trouble.
When you strip away Huang's half-hearted attempts at spin doctoring, the CNET interview is an admission of failure for Tegra and Nvidia's efforts to get into the ARM processor business. Total production of automobiles for 2013 was 65.4 million, according to the International Organization of Motor Vehicle Manufacturers, compared to 1 billion smartphone shipments (according to IDC) and 195 million tablets (according to Gartner). Automotive is only a market you chase when you've been chased out of the mobile market.
Nvidia has been bucking two important trends that work against it in the long term. The first trend is the incorporation of graphics processing into computing processors, eliminating the need for a discreet graphics processing chip. Many Intel processors and virtually all ARM processors have built in graphics capability, making them "systems on chip" (SOCs). As process nodes shrink, more silicon real estate is available for graphics processors, and the on-board graphics capabilities of SOCs continue to improve.
AMD's (NASDAQ:AMD) "accelerated processing units" used for the latest generation of game consoles from Microsoft (NASDAQ:MSFT) and Sony (NYSE:SNE) demonstrate the power of combining conventional CPU and GPU on a single chip. The trend will only continue. Eventually, except for a very limited set of applications, all computing devices will employ SOCs, from the server to the desktop to the smartphone.
So far, Nvidia has been successful bucking this trend. Its high performance graphics cards sell well for PC gaming rigs and for professional applications where the graphics processing power of Nvidia's GPUs is needed. Although GPU sales grew in 2013, I doubt the trend will continue. The increasing importance of mobile devices, the success of the new game consoles, and the move by Intel and the ARM industry to 14-16 nm class process nodes for the next generation of SOCs will begin to reduce discrete GPU sales by the end of 2014.
The other trend that Nvidia has been bucking has been the paradigm shift away from commodity production of processors by a traditional supplier such as Intel toward design of custom SOCs by device makers such as Apple and Samsung with production performed by silicon "foundries" such as TSMC. When Texas Instruments (NASDAQ:TXN) withdrew from the ARM mobile processor market (also to pursue automotive and other embedded applications) in November 2012, it specifically pointed to this trend as the reason for its action.
Even as Intel and AMD pivot away from the commodity model to more custom work, there's still a small but shrinking assemblage of companies fighting it out to be ARM processor suppliers, including Qualcomm (NASDAQ:QCOM) and Nvidia. Here, the usual approach is to offer some value-adding discriminator. Nvidia's discriminator was to be its superior graphics capability.
Blind Sided by 64 Bit
Superior graphics as a discriminator might have worked except that right around the time the delayed Tegra 4 was appearing in the new Surface 2, Apple released its new phones and tablets based on its 64 bit A7 SOC. At the time, there was a lot of misinformation circulating about the significance of the A7, with many claiming that the 64 bit capability would have little immediate impact on performance. These claims were crushed decisively by Anandtech's review of the iPad Air, which verified Apple's claims of roughly a 2X performance improvement.
Due to Apple's penchant for secrecy, the ARM processor industry was blind-sided by the A7 and left scrambling to come up with 64 bit processors in response. Holiday sales of Microsoft tablets were dismal at an estimated 1.275 million units in 2013 Q4, and this includes sales of the Intel-based Surface Pro.
With the A7 in iPad and iPhone, Apple set the bar much higher for "premium" mobile devices. The Tegra 4 had become a commodity processor with a premium price. As Huang admits in the CNET interview, that was not a competitive combination.
The industry probably shouldn't have been blindsided by the A7. The A7 was primarily a defensive move intended to counter the arrival of Intel's Bay Trail SOCs, which were also highly capable 64 bit processors. Intel had published its road map for Bay Trail a year in advance, so the industry had plenty of warning.
Bay Trail started arriving in Intel Based Android tablets at the end of 2013, but probably didn't gain much market traction until Intel implemented its "contra-revenue" pricing plan that made Bay Trail more price competitive with ARM mobile processors. As I pointed out after Intel's Q1 earnings report, contra-revenue negatively impacted Intel's mobile device revenue, but it almost certainly stimulated sales of Bay Trail. Android tablet makers now had a higher-performance 64 bit alternative to the Tegra 4.
The most important take-away from the CNET interview is that there's absolutely no mention of a 64 bit Tegra. In order to stay competitive in mobile, Nvidia would have had to be able to offer a 64 bit version of its own. The decision to withdraw from mobile says that Huang realized that it was already too late to start a 64 bit processor for the mobile market.
If Microsoft intends to continue with its ARM-based Surface RT tablets, which in itself is doubtful, it would almost certainly want a 64 bit processor in order to maintain its position as a premium tablet alternative to Android and iOS. This probably means that Nvidia was no longer in the running for Surface RT, which would have been the final blow to Nvidia's mobile ambitions.
Nvidia's next earnings report due in August should start to show the impact of its mobile failure, but it may be difficult to discern. In the year ago quarter, Tegra processor revenue declined sharply to about $50 million with the delay of the Tegra 4 and the flop that was the first Tegra 3 based Surface RT. Next to that, almost any results for Tegra for this quarter will look better. It may require several quarters for the impact of Nvidia's withdrawal from mobile to become apparent. By Q4 of this year, I expect Tegra revenue to be less than half of what it was for 2013, or about $50 million.
Disclosure: The author has no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.