This week Intel announced its mobile discrete GPU strategy, ARC, and, as an ex-Competitive Analyst, I find it fascinating. I have not had a chance to review the part myself, but on paper the specs are impressive with equal to or better parity to AMD and NVIDIA and a potential advantage in upscaling content. This could be a godsend when you, like I do, play older games. But that typically is not enough to move against a market dominated by two vendors any more than it was effective in taking CPU dominance from Intel. So, what is interesting is Intel’s first move is in the mobile sector where it has a significant capability to lock out the other two vendors. Intel is dominant in integrated graphics and can provide a cheaper, more elegant upgrade path—at least on Intel CPU-based laptops—than AMD or NIVIDIA can.
Let’s explore Intel’s strategic approach to mobile graphics this week.
Overcoming a dominant vendor
Typically, when there are one or more dominant players in a market, government intervention is required to open competition. A challenger, no matter how big, will be unable to catch up with the dominant companies unless they have cut back development and are just milking the market. RCA, Shell Oil and AT&T were opened by the government. IBM remained unaffected until it turned the mainframe into a cash cow in the 1980s and almost went under.
But these paths all require either a mistake, which you cannot rely on, or government intervention, which is a dual-edged sword. For instance, Google is now paying the price for using the European Commission to open Microsoft’s interfaces because that same commission, after being enabled by Google, Sun and Oracle, is now looking at Google (and has gone after Oracle in the past).
The method with less risk and less reliance on luck is to first create a beachhead either by tightly focusing on an area that is not a high priority of the larger vendor (much like AMD focused on desktop computing a while back against Intel) or where you already have a significant advantage.
This is the strategic path Intel has taken. It’s going after mobile first because, in that sector, it has a solid advantage due to the nature of how laptops are built (without graphics cards). So, Intel can, with far less effort, potentially get to critical mass.
Why critical mass or a trend toward it is important
GPUs require substantial third-party software support. Microsoft support is a given, but you also need support from the software companies the user chooses. Without that support, even with better specs, you will underperform. The software will not call the advanced features that drive the extra performance. This is a cart-before-the-horse problem in that, if you do not have the support, users will not buy the product. If you do not have enough users, the developers will not support you. But given Intel’s massive dominance in laptops, developers will assume success and are far more likely to support the technology early.
Thus, the path to market Intel has chosen is likely to be more successful by far than if it had first tried to produce a graphics card and then likely underperformed in sales, resulting in either a very long ramp to volume or a failed offering.
It is ironic that Intel finds itself where AMD and Qualcomm are with respect to Intel’s CPUs against AMD and NVIDIA with GPUs. Intel is rightly first focusing on a platform that should, assuming execution and the parts are as good as presented (as I expect they will be given leadership jobs depend on success), create the impression that Intel will reach critical mass long before they do. This should motivate developers and eventually provide enough support for Intel to move more broadly into the market and turn what is now a two-horse race on graphics into a three-horse race.
Now the question remains: can Intel keep up? The GPU market moves more quickly than the CPU market does today, and both AMD and NVIDIA have been going flat out for a while now. This is only the first engagement, but the odds, again assuming Intel executes, favor the path it has taken over far more risky and uncertain alternatives. It will be an interesting GPU year.