Image credit: NVIDIA.
Graphics specialist NVIDIA (NVDA 3.65%) is releasing new graphics processors based on its Pascal architecture at a furious pace. On Monday , NVIDIA apparently announced (via Hardware.info) two new graphics cards aimed at the professional visualization market.
The first is a product known as the Quadro P5000. This card is reportedly based on the same GP104 chip that powers the gaming-oriented GeForce GTX 1080, but with 16 gigabytes of GDDR5X onboard, rather than the "mere" 8 gigabytes found on the GTX 1080.
The second is the Quadro P6000. Hardware.info says this is based on NVIDIA's GP100, but given that it packs 24-gigabytes of GDDR5X, I have to agree with VideoCardz.com in thinking that this is actually the GP102 that powers the recently released prosumer-oriented NVIDIA Titan X card.
Interestingly, unlike the Titan X, the Quadro P6000 apparently has all 3840 CUDA cores on the graphics card enabled (the Titan X only has 3584 CUDA cores enabled).
Hardware.info says that these cards will be available in the October timeframe.
NVIDIA's lucrative (but growth-challenged) business
Professional visualization is a segment that's quite material to the company's revenue base. In NVIDIA's fiscal 2016 (which ended Jan. 31), this segment brought in $750 million in revenue, approximately 15% of its overall revenue.
This segment also carries quite high gross profit margins; the hardware that NVIDIA sells here is very similar to the consumer grade stuff that it sells to gamers. However, the products sold under the Quadro branding are much more expensive than their GeForce counterparts as they come with firmware/drivers specifically tailored to the requirements of professional visualization applications.
Unfortunately, since fiscal 2013, revenues in this segment have been practically flat ($719 million in fiscal year 2013; $750 million in fiscal year 2016). Given that this market is quite lucrative for the graphics specialist, it's not surprising to see the company continue to try to push newer and better products here.
These new cards look like solid improvements over the prior Maxwell-architecture based Quadro M5000 and M6000 offerings, respectively. However, I suspect that the financial performance of NVIDIA's professional visualization segment will be largely driven by market trends rather than by the company's product offerings.
Nice operating leverage
The nice thing about NVIDIA using the same hardware for both its gaming and professional visualization products is that it provides it with quite a lot of operating leverage. However, the Quadro revenues don't come entirely free.
NVIDIA says that it does "joint development with [professional] applications providers and rigorous in-house testing" to certify compatibility of its Quadro products with "all industry leading applications."
This certification takes time, effort, and -- ultimately -- money, so there are incremental operating expenses associated with getting these products to market. These products are also relatively low volume compared to, say, gaming products.
To make these incremental operating expenses worthwhile, the company charges more for comparable Quadro hardware relative to the gaming-oriented GeForce hardware. The high gross margins ultimately make up for the incremental operating expenses associated with this segment.
Note that since NVIDIA is committed to this market, and is going to spend on development anyway (operating expenses shouldn't scale much with revenue), any incremental revenue gains from this segment can potentially have a really nice impact on the bottom line.
We'll see in the coming years if sustainable growth in this segment is a realistic expectation, or if it will simply be a flattish-revenue cash machine going forward.
