There is no question that Apple (AAPL 1.68%) was one of the first mobile device vendors to really focus on graphics performance in its iOS devices. Further, Apple is well-known for pushing its Mac silicon supplier, Intel, pretty heavily on the development of high-performance integrated graphics. Apple knows that gaming is a major use case for mobile computing and as a result has typically optimized its system-on-chip products to offer great graphics performance.
The early days: using off-the-shelf Imagination IP
Imagination Technologies (IMG +0.00%) is well-known as a premier graphics IP vendor for many mobile system-on-chip products. Its graphics IP is very power efficient and offers rather excellent performance per unit of area. For years, Apple has employed graphics IP blocks designed by Imagination for its "A-series" system-on-chip products and, frankly, the results have been quite good.
This led to a major run in Imagination's stock following the release of the iPhone and, as the smartphone market in general proliferated, Imagination found a seemingly endless stream of customers for its IP. This gigantic demand allowed Imagination to beef up its R&D efforts significantly and has, in general, been a virtuous cycle for the company (although recently competitive threats from ARM, Vivante, and NVIDIA have begun to worry investors).
The future: in-house GPUs
A quick look at Apple's job boards shows that the company is very aggressively staffing up an internal GPU IP development team:
(Source: Apple.com)
It is likely that Apple will be very aggressive in designing graphics processors that are custom-tailored to the needs of its iOS products. That's not to say that Apple doesn't have a lot of leverage over at Imagination Technologies (Apple is a 9.5% shareholder of Imagination, so there's plenty of leverage), but for a major company like Apple with a world-class silicon team, using off-the-shelf IP simply isn't optimal when Apple could truly optimize the silicon to meet its exact needs (and Apple controls the entire software stack, meaning it intimately understands these needs).
The rest of the world is moving fast – Apple needs to lead
Look at Qualcomm, Intel, and NVIDIA – three merchant chip vendors that are very rapidly iterating their in-house GPU IPs at the high end. In order for Apple to continue to justify doing its own chips (and not simply caving in and buying off-the-shelf chips from the merchant vendors), it will need to develop chips that – for the products that Apple is building – are superior to what the merchant vendors are doing.
This is going to be tough. Intel, for example, is throwing gobs of R&D money at low power CPUs, next generation GPU architectures, modems, and so on. Qualcomm and NVIDIA have already proven that they can put out some pretty slick hardware with their respective Snapdragon 800/805 and Tegra 4/K1 chips. Apple needs to do better than these companies that live and breathe chips in order to "win".
Licensing off-the-shelf GPU IP at the rate that Imagination can deliver it simply won't cut it. Apple's in-house custom GPU is likely to make trade-offs much more appropriate for a very high margin, performance-oriented smartphone/tablet than trade-offs made in a "one-size-fits-all" type of deal. For low end to mid-range SoCs, off-the-shelf IP will be just fine and dandy, but for a premium product, likely not.
Foolish bottom line
Whether it's in the iPhone 6, 6s, or 7, Apple is likely to make a move to an in-house GPU architecture, likely aimed at performance levels that exceed what the current off-the-shelf IP offers today. Apple is likely to make its GPUs bigger than most, but given how much money it makes per iPhone/iPad, the higher die cost is likely worth it for the sake of leadership performance/user experience. Yet another example of Apple's innovation engine at work –even if it's not obvious to most consumers.
