NVIDIA had a record quarter, but what got them there?
It always feels a little odd when covering NVIDIA’s quarterly earnings due to how they present their financial calendar. No, we are not reporting from the future. Yes, it can be confusing when comparing results and getting your dates mixed up. Regardless of the date before the earnings, NVIDIA did exceptionally well in a quarter that is typically the second weakest after Q1.
NVIDIA reported revenue of $1.43 billion. This is a jump from an already strong Q1 where they took in $1.30 billion. Compare this to the $1.027 billion of its competitor AMD who also provides CPUs as well as GPUs. NVIDIA sold a lot of GPUs as well as other products. Their primary money makers were the consumer space GPUs and the professional and compute markets where they have a virtual stranglehold on at the moment. The company’s GAAP net income is a very respectable $253 million.
The release of the latest Pascal based GPUs were the primary mover for the gains for this latest quarter. AMD has had a hard time competing with NVIDIA for marketshare. The older Maxwell based chips performed well against the entire line of AMD offerings and typically did so with better power and heat characteristics. Even though the GTX 970 was somewhat limited in its memory configuration as compared to the AMD products (3.5 GB + .5 GB vs. a full 4 GB implementation) it was a top seller in its class. The same could be said for the products up and down the stack.
Pascal was released at the end of May, but the company had been shipping chips to its partners as well as creating the “Founder’s Edition” models to its exacting specifications. These were strong sellers throughout the end of May until the end of the quarter. NVIDIA recently unveiled their latest Pascal based Quadro cards, but we do not know how much of an impact those have had on this quarter. NVIDIA has also been shipping, in very limited quantities, the Tesla P100 based units to select customers and outfits.
The combination of solid Maxwell/900 series sales combined with the first of the next generation GPUs that NVIDIA has introduced provided a very strong quarter. Automotive is a growth area for the company and one they have been excited about for years. Throw in the massive push towards deep learning and NVIDIA certainly looks to have a steady stream of income for years to come from these technical investments.
The company has also invested a lot of time and development into addressing the VR market. The future of this market looks bright, but so far the cost of entry is very high for acquiring an Oculus or Vive product. Software support is slowly growing and we see much more inertia from the market and developers than we have in the past. We can compare this directly with the 3D panel market for both the PC and consumer markets. While many monitors and TVs supported 3D, it ended up being more of a checkmark than must-have technology. We expect VR to succeed where 3D panels have failed, but that result is not set in stone. NVIDIA’s technologies surrounding VR all add to the experience while cutting down or smoothing out the negatives. They also have the fastest single GPU cards available that would be the target for those buying the VR products. These investments have not quite paid out yet, but they set a solid foundation of software and support for the implementation of VR in a variety of applications and games.
Manufacturing Going Forward
NVIDIA was first to market with a new generation of large, standalone GPUs based on the latest process technology from any of the pure play foundries. While the cellphone SOC makers have been leaning on the “early” and “high performance” process nodes with Mali, Adreno, and PowerVR units, this is the first we have seen from the big GPU manufacturers.
Samsung and TSMC both have flavors of FinFET process nodes that differ in some significant ways. Samsung’s process is physically smaller than what TSMC offers, but in terms of electrical properties they in theory should be similar. Apple has tapped both foundries for products, and basic results under test conditions have shown them to be fairly similar with Samsung having a slight edge in feature size.
Samsung already showing off 10 nm wafers. Could this be the future foundry partner of choice for NVIDIA? (image courtesy of SemiWiki)
It is not simple to port a design from one process to another. NVIDIA looks to have taken Maxwell, added more functional units without drastically redesigning the architecture, and applying it to TSMC’s 16nm FF+ process. This is still a complex procedure. It seems that NVIDIA is applying a “tic-toc” style of production here with taking a known architecture and applying a new process node to it rather than creating a new architecture on a new process. When dealing with a product that is comprised of 8 billion transistors, removing as many variables as possible to achieve a working product is simply good business.
NVIDIA has a long and productive relationship with TSMC and they have been their primary foundry partner since the Riva TNT. Times seem to be changing though. TSMC’s 16nm FF+ is still a very new process that is constrained currently. We have not heard detailed information about exact yields, design issues, or late changes to libraries that would affect the products in question. We also know that Microsoft has placed significant orders with TSMC for their 16nm FF+ based XBox One S SOC. It is surprising to find out that NVIDIA has entered into an agreement with Samsung to utilize their 14nm FFP process for future chips. We do not know the full details here, but it seems as if NVIDIA is playing it safe by leveraging a second supplier. Perhaps the relative scarcity of the high end GTX 1080 over the past few months had just as much to do with TSMC able to deliver product as much as demand?
This leads to other questions. Why did NVIDIA seek out Samsung rather than attempting a contract with GLOBALFOUNDRIES? Samsung has been doing brisk business with other companies while it is pretty well known that GF could certainly use more customers to fill up wafer orders and fab utilization. AMD is utilizing GF for their latest Polaris parts, but many questions surround the actual implementation that we are seeing. Polaris parts run about the same in terms of efficiency as the previous generation Maxwell parts, all the while being clocked well below what we are seeing with NVIDIA’s latest Pascal chips. Is this due to design, or due to issues with GF’s version of the Samsung 14nm LPP. If GF is still getting up to speed with 14 LPP, then it is understandable that NVIDIA would go straight to the source and deal directly with Samsung. Also of note is that Samsung will likely get a 10nm process up and running before we see it used by GLOBALFOUNDRIES. This is merely speculation, but these questions do come to mind when looking at the overall landscape of the industry.
The Next Quarter
NVIDIA is bullish about their upcoming quarter. They are selling their next generation parts at a very brisk pace. Reviews for these cards have been overwhelmingly positive and they compare very well to the latest offerings from AMD. NVIDIA is also benefiting by moving good volumes of higher end products that support higher margins than those found in the $200 to $250 range. While AMD is pushing out solid numbers of products in the $250 and below market, margins will be much slimmer there.
Some may call it… the moneymaker.
The current estimate of what Q3 2017 will bring is in the area of $1.68 billion. If I am not mistaken, this will be the highest revenue recorded for NVIDIA in any quarter over the lifespan of the company. While it is too early to get excited, this could lead to another record once Q4 is completed. NVIDIA is counting on continued strong sales of their GPUs across the line, their refreshed Quadro cards, and the automotive and deep learning enterprises. Potentially diversifying foundry suppliers could lead to more available product to put on shelves for the important Holiday buying season.
NVIDIA has done very well over the past couple of years and through some rough patches. They continue to move forward with design and are quite agile in addressing new markets and abandoning areas that were not profitable. They seem to remain focused and have been able to capitalize on the mistakes of their competition. We shall see if some of their new endeavors actually make money for the company, but for now their current products are propelling the company forward.