Over on Twitter (NYSE:TWTR), Rys Sommefeldt, Editor-in-Chief of popular 3D graphics-oriented website Beyond3D and Business Development Manager at graphics IP vendor Imagination Technologies (NASDAQOTH:IGNMF), made the following very interesting "tweet":
From a good look at Gen9, I now bet that Intel will get back into the discrete PCIe graphics card business with a Gen10 variant— Rys Sommefeldt (@ryszu) September 2, 2015
In other words, he thinks that Intel (NASDAQ:INTC) -- which currently only offers its graphics technology integrated onto its CPUs -- could use its next generation graphics processor design to enter the market for stand-alone graphics chips. Such a move would put Intel into direct competition with NVIDIA (NASDAQ:NVDA) and Advanced Micro Devices (NASDAQ:AMD).
Although Sommefeldt makes the interesting point that there could be money to be made for Intel in this market, I don't think Intel will actually go ahead and do that. Here's why.
How much of the market would really be open to Intel?
If Intel were to enter the market for discrete graphics cards, it would -- as Sommefeldt suggests in a subsequent "tweet" -- probably make a play for the mid-range of the market.
I'd argue that the business case for trying to compete broadly in the market for stand-alone graphics is a bit iffy. I'd estimate that the market for such chips is probably in the ballpark of $4 billion, with NVIDIA capturing the majority of it and AMD as a distant second.
If Intel were to enter the market, it would have to expend not-so-insignificant resources into trying to get its piece of a, perhaps, $4 billion pie. For a company that's on track to generate over $55 billion in sales this year, I'm not sure that this is all that attractive.
The already-iffy business case for Intel to enter this market becomes even less attractive if Intel doesn't try to go for the high-end of the market, where a significant portion of the money to be made from the discrete graphics processor market lies.
Although one could argue that Intel might be willing to make the incremental investments required to go after this market, the risk-to-reward ratio of trying to go up against NVIDIA and AMD in this segment -- which have established gamer-focused brands (GeForce and Radeon, respectively) and long histories of developing high-performance graphics chips -- seems rather poor.
Intel's strategy seems a better way to create shareholder value
Intel's current strategy with integrated graphics is an interesting one. For a given class of product, Intel tends to offer two levels of graphics performance.
There's the baseline level of graphics present in most of its chips which come with reasonable levels of 3D performance as well as the company's latest multimedia features. In a modern PC, integrated graphics is a must, so Intel seems to try to make even this "baseline" level of graphics fairly competitive.
Where things get really interesting, though, is what it's doing with its higher-end variants. Intel seems to be trying to -- in some cases significantly -- increase the performance of its higher-end Iris/Iris Pro integrated graphics engines with each generation.
Intel's apparent goal is to try to displace low-end/mid-range discrete graphics chips, particularly in mobile form factors, by trying to offer similar performance as those parts and capturing the value that would have otherwise gone to the discrete graphics chip vendor.
Ultimately, Intel's goal here is to try to improve the average selling prices of its chips, which is probably the most sensible way for Intel to try to capitalize on more powerful graphics solutions.