With the bowl of punch nearly gone, miscellaneous party favors littering the floor, and S|A writers stumbling about in various states of sobriety, it is time to settle down and discuss the results of Nvidia’s recently concluded Q2, FY 2011 earnings conference call. Much of the information we already knew or surmised after the lowered revenue guidance bombshell they dropped on us in late July, but half the fun of these kinds of calls are the molehills-turned-mountain that companies create to keep investor’s wallets open. Let’s start hiking.
In Q2 of fiscal year 2011, Nvidia reported revenues of $811.2 million and realized a GAAP net loss of $141 million, compared to a reported GAAP net profit of $137.6 million just a quarter ago. At 16.6%, gross margins were also down both sequentially and year-over-year when compared to 45.6% from last quarter, and 20.2% a year ago. If you toss GAAP out the nearest airlock however, the numbers change to reflect a realized net income of $20.1 million for the most recent quarter. This is due largely to continued bumpgate write-offs, or as they officially put it in their GAAP vs. Non-GAAP reconciliation, “Net charge arising from a weak die/packaging material set.” The total reported write off amounted to $193.9 million in Q2.
The official outlook offered by Nvidia for Q3, FY 2011 indicates expected revenue increases of 3-5% over that of Q2, which would put revenue estimates at $835.5 – $851.8 million for Q3. If true, this would still be significantly down (6% – 8%) from the $903.2 million in revenue they reported in Q3 2010, and would even fall below reported revenue figures from Q3 2009 ($897.7M), and 2008 ($1.12B) as well. Nvidia is expecting gross margins to increase to 46.5-47.5% in Q3.
According to Nvidia, the driving forces behind the expected Q3 increases are as follows (in no particular order): StarCraft II, Fermi-based Quadro, StarCraft II, Fermi-based Tesla, StarCraft II, blind faith in Tegra based products, StarCraft II, and ramping of Fermi rev 2 derivative cards (GF104, 106, & 108) presumably for playing StarCraft II. Now the more astute readers may have noticed a subtle pattern taking shape in the preceding list. That’s right, they have indicated that their product mix will be converted to primarily Fermi based chips across their product portfolio. While no exact timetables were furnished for the budget/midrange Fermi rollout, it was alluded to that these parts would be available soon, presumably when JHH takes a break from his StarCraft II ladder matches and pushes the big, red, “Ship” button on his desk. Here’s to hoping somebody reaper-rushes captain Huang soon so he can get product on shelves before schools start back up in a couple weeks.
In all seriousness now, StarCraft II was mentioned ad nauseam throughout the whole conference call. It started out being used as an example of great games driving sales of PC graphics hardware. Fair enough. From there they proceeded to whip the StarCraft II horse at every opportunity, as though hypnotized (or paid) by some mystical marketing guru at Blizzard. Here’s the rub: StarCraft II will run on a mother loving 6600GT or 9800pro (7 years old), and even the recommended requirements are only at 8800GTX/HD 3870 levels (3-4 years old) . Most PC gamers will already have enough hardware to amply run StarCraft II, so the upgrade market doesn’t seem particularly lucrative based on that single title (Especially considering that StarCraft II is a DX9 game). There are plenty of other current/upcoming releases (For example Crysis 2, or AvP which makes use of tesselation) that would have better pimped out the “graphics leader” image they want to project. The question now arises, was this merely an embarrassingly executed marketing gimmick, or a subtle calculated hint (for better or worse) at the capabilities of their upcoming budget lineup?
Back from that tangent, Nvidia actually admitted that their consumer GeForce division is currently in the tank multiple times during the call, but deflected questions and forecasts towards other segments they are counting on to pick up the slack. Quadro professional cards, in particular, are at the forefront of Nvidia’s mind right now. Recent reviews have shown them to be quite potent in professional applications (as you might expect, since Fermi was essentially designed for this market, not gaming efficiency), and Nvidia is banking on pros upgrading to the newer models to reap the rewards of a claimed 5x-8x performance boost. If the performance numbers are to be believed, these new Quadros could prove quite popular, and the high average selling price of these cards might finally squeeze a profit out of the ginormous GF100 die.
Tesla was also touted as having another record quarter, with a bright future ahead of it. This is not at all surprising considering Tesla is still a relatively new product branch, and sales have never been incredibly good to begin with. During the call Nvidia revived rumors of working on an Exa-Scale, GPU-based supercomputer in conjunction with Oak Ridge National Laboratory and several “top universities”.
In Tegra news, Nvidia initially chatted up the SoC platform as practically being the second coming, even though to date their major design wins for the beleaguered chip can be counted on one hand. Amusingly, when asked whether Tegra would be coming soon in smartphone or tablet form, Nvidia replied: “We hope so.” They went on to state that they have no idea which will ship first [smartphone or tablet] but that whatever it is will be “pretty amazing.” They claim that Tegra sales are expected to generate enough revenue in the (near?) future to make up for lost chipset business. There was no mention of Tegra 3 during the call, but given the lack of attention that people have given T2, it would make sense to have something more appealing in the works. If at first you don’t meet your power budget, try, try again.
Just about everybody knows that Nvidia’s chipset business is bust these days. During the conference call this bit of common knowledge was officially confirmed with JHH himself stating that their chipset business is predominantly Apple platforms. If you’ve paid attention to the latest round of Macintosh refreshes, you’ll undoubtedly know that Apple has kicked Nvidia out of their recent product lineups in favor of Intel Core i-Somethingoranother CPU’s with integrated graphics, and Radeon 4000/5000 series chips to pick up the slack. This means that Nvidia essentially has no chipset business, and for once they admitted it by stating they expect chipset shipments to be only 25% of what they currently are over the course of the next two quarters.
Despite the seemingly weak results and less than exciting forecasts, Nvidia shares rose somewhat in afterhours trading following the call. For all intents and purposes their consumer business is FUBAR in the near term. The launch of GTX 460, while met with critical acclaim, is not going to be a huge money maker due to its comparatively low ASP, and large (albeit smaller than GF100) die-size. Even if GF106, and GF108 follow similar price/performance patterns and get released in some reasonable time frame, (read: yesterday) they will still have to deal with the fact that ATI’s competing parts will have the advantage of time-to-market (and thusly OEM adoption), better power draw characteristics, smaller dies and more price wiggle room as a result. The potential of a Southern Islands (HD6K) refresh further complicates matters, particularly if the refresh goes top-to-bottom throughout ATI’s lineup as smoothly as the 5000 series seemingly went.
As a result Nvidia’s crutch lies in the professional markets where they can command higher prices until such time that the consumer products are once again able to compete (profitably) across the board. One thing to contemplate is the utter lack of any substantial GeForce 500 series talk. GF104, 106 & 108 are being sold as “Fermi 2” according to the higher-ups at Nvidia, however they are still parts meant to fill in the blanks of the 400 series lineup. We have had rumors and confirmations of ATI’s upcoming projects (on both 40nm and 28nm) for a long time now, but things are eerily quiet on the green side of the fence. Barring another round (or three) of re-badging gimmicks, which they’d never dream of doing *cough*, 500 series parts wouldn’t come until well into 2011 when TSMC’s 28nm process comes online. Perhaps they are keeping quiet until their low end current generation products finally launch, or perhaps they simply have nothing to talk about at this point.
We’ll reserve final judgement on the consumer graphics situation until Nvidia’s final budget and midrange products start shipping. It will not be an issue of price/performance, but rather a complex mixture of efficiency, OEM acceptance, HD6K anticipation (plus subsequent HD5K price drops) and profitability of the lower end chips (which will have larger dies than their competitors) that will make or break Nvidia in the consumer segment this holiday season. Needless to say, at face value, based on these variables we should be seeing a large increase in coffee CAPEX to keep Nvidia’s marketing team highly caffeinated over the next few quarters. S|A