Nvidia announced its quarterly earnings this week, and the company posted record-breaking numbers in just about every category. Revenue leapt 54% to just over $2 billion, the first time Nvidia has had a $2 billion quarter. Its gross margin set a record at 59%, with a record 63% increase in gaming-derived revenue. There’s no doubt about it: Nvidia’s dominance of the high-end GPU market is allowing it to suck down all of the spending that would normally have been at least somewhat split between itself and AMD.
Gaming was the big revenue booster for Nvidia. Other sources, like Quadro, grew a modest 9%, but nothing like the leap that gaming GPUs saw. Data center revenue grew by 59% as well, though this was much smaller in absolute terms ($230 million in data center sales, versus over $1 billion in gaming revenue). Jen-Hsun Huang, the CEO of Nvidia, noted that he saw strong growth in AI, though he opted not to break those figures out at this time. Nvidia’s automotive program is also going well, with $127 million in revenue (a 61% increase year-on-year) and a 7% increase sequentially.
Overall, Nvidia seems to have found niches that fit it much better than the smartphone and tablet market. While the company enjoyed some solid success with Tegra 2, Tegra 3 wasn’t a huge hit and the uptake for Tegra 4 and its following products was very small. The Shield K1 is a great Android tablet (even if it’s getting a bit long in the tooth). What Jen-Hsun talked about in the conference call is how Nvidia wants to build a computing platform that stretches from GPUs on the desktop, to cloud solutions like GRID, to automotive computing and self-driving cars, to HPC. It’s not that the rhetoric is different, but the fact that Nvidia is well on its way to doing it.
There are, of course, plenty of questions still to be answered. No one knows how quickly self-driving cars will enter the market, AI and deep learning are still in their infancy, and the majority of computer systems built for HPC – at least, those on the TOP500 – still don’t use accelerators of any kind to improve performance. Still, Nvidia is performing exceptionally well.
Any discussion of these trends inevitably raises the question of how and where AMD will offer competitive performance. Right now, we’re in the doldrums as far as competing directly with Nvidia’s high-end. We’ve known this would be the case until Vega launches, and I don’t think AMD is somehow critically behind simply because it’ll launch later than its competitor. Fermi, after all, debuted nearly a year after the HD 5000 family. It also took AMD nearly a year to build a decent competitor to Nvidia’s G80 back in 2006, and yet by 2008 the HD 4000 series was offering strong competition against Nvidia.
What Nvidia has done, that AMD doesn’t seem to have an answer for, is differentiate its GPUs into a variety of markets. OpenCL and HSA may be great on paper, but uptake has been nil. Three years after promising a version of Java that supported HSA and talking up multiple games and applications that would take advantage of Kaveri’s benefits, the number of HSA apps in the market is very nearly zero.
AMD’s Boltzmann initiative from last year was interesting and may be winning some small support, but the company needs to launch both Zen and Vega in the next few months. Based on their mutual performance, we’ll have a much better idea how well-positioned AMD is to slug things out in the CPU and GPU market. AMD’s console wins are what’s keeping the company alive these days, but it needs to demonstrate stronger performance in PC gaming or other segments if it wants to take on Team Green across the entire computer industry.
What’s particularly ironic about this, of course, is that after AMD announced that its future CPUs would actually integrate graphics chips, plenty of people thought this could be the end of Nvidia as a GPU manufacturer. Clearly, that’s not what happened. And while yes, there have been controversies in the gaming community about programs like GameWorks, Nvidia has done an excellent job winning business and executing its strategies in markets that have nothing to do with gaming or game performance.