Back in November, Intel (INTC -9.33%) announced that it had hired Raja Koduri, a respected figure in the world of high-performance graphics processors, to run its newly formed Core and Visual Computing Group.
The company said in that announcement that it intends to "expand [its] strategy to deliver high-end, discrete graphics solutions."
Image source: Intel.
By way of background, virtually every computer that's made today -- whether it's a bulky desktop PC, a sleek notebook computer, or even a smartphone -- has some sort of graphics processor built into it. By far the most common way that graphics processors are deployed is through what is known as integrated graphics.
All this means is that the graphics processor isn't a stand-alone component inside of the computer -- instead, it's integrated into the same chip as the other key processing components in the system (like the CPU, memory controller, and so on).
However, the performance of such integrated solutions is ultimately limited by space, cost, and power constraints. So, for applications that demand more graphics performance (for example, running the most graphically demanding 3D video games), it makes sense to build graphics processors that are independent of any other components. These are called discrete graphics processors.
This is the market that Intel will be re-entering.
Some have suggested that since Intel will be a challenger in this market, it will need to price its chips fairly aggressively to have any hope of gaining traction. Moreover, since those chips will be priced aggressively, the profit margins per chip sold will be lower than the margins of, say, Intel's CPU products.
The thinking, then, is that Intel might view the opportunity cost of building lower-margin graphics products compared to, say, using its manufacturing capacity to build higher-margin PC processors or data center chips as too high.
Here's why I don't think this won't be an issue.
Intel can do both
The first thing that's important to consider is that the stand-alone graphics processor market is entirely greenfield for Intel -- the company has precisely zero market segment share here today and it'll stay that way until it releases its first products. So, as long as Intel sells these stand-alone graphics processors at a positive gross profit margin (in other words, the selling prices of those chips are higher than Intel's manufacturing costs), Intel is doing OK.
Now, to be clear, selling products that carry lower gross margin percentages than the company average gross margin percentage necessarily dilutes the average. However, total gross profit dollars will go up, which ultimately means fatter profits for stockholders -- a big win.
Image source: Intel.
The next thing to consider is that the opportunity cost trade-off would only really happen if Intel were manufacturing capacity constrained and was forced to make a choice between producing enough of its higher-margin products to meet demand or producing the lower-margin graphics processors.
This scenario isn't terribly likely to occur because chip manufacturing companies generally put in capacity based on the overall demand that they expect. So, if Intel expects that there will be serious demand for its graphics processors (something that should become clear to Intel from conversations with its major customers), then it should put enough manufacturing capacity in place to handle all of its expected needs.
Moreover, Intel has made it clear that it tends to build in a little extra capacity compared to its expected demand to allow it to profit from better-than-expected demand should that situation arise. That additional capacity, should it go unused, does hurt profitability a little bit, but it's a worthwhile trade-off for the flexibility that it brings and the peace of mind that it could afford Intel's major customers.
Only time will tell if Intel's efforts in the stand-alone graphics processor market succeed, but I don't think lower-than-corporate gross profit margins on the chips should be a gating factor to Intel's efforts here.
