Apple’s Semiconductor Prowess

Apple has been taking great strides to control and design more of their own proprietary semiconductor components. One piece many of us believed was on the horizon was the GPU. Apple is a stakeholder in Imagination Technologies and has used their GPU architecture for some time. Earlier today, Imagination sent out a press release informing the world, and specifically their investors, that Apple will no longer be licensing their GPU IP and will instead be going to market with their own proprietary GPU. There are a few interesting angles to explore with this new information.

First, some have suggested this is a ploy by Apple to drive the stock price down of Imagination Tech in order to then pursue a buyout. While there is certainly a chance of this clever strategic move, I tend to think it ranks much lower on the list of likely scenarios. One main reason is it’s risky and there is no guarantee the strategy works when you have a number of other tech companies like Microft, Google, Amazon, etc., who would also be very interested in pursuing a buyout — Apple would not be the only one at the negotiating table.

Imagination wanted to get out in front of this situation with the press release and worded it quite specifically to attempt to appease investor concerns. I read a lot of buy-side analyst commentary on Imagination and this narrative that Apple would drop them as an IP licenser was foremost among investor concern. Imagination knew this and that losing Apple would negatively impact their stock. That is indeed what happened.

This part of the Imagination statement is one worth digging into:

Apple has not presented any evidence to substantiate its assertion that it will no longer require Imagination’s technology, without violating Imagination’s patents, intellectual property and confidential information. This evidence has been requested by Imagination but Apple has declined to provide it.

Further, Imagination believes that it would be extremely challenging to design a brand new GPU architecture from basics without infringing its intellectual property rights, accordingly Imagination does not accept Apple’s assertions.

Imagination wants to give investors a sense that, even if Apple is going down this route, they will ultimatlely still need some Imagintion IP in the end. It is certainly true that building a GPU from the ground up will likely require someone’s IP in some way for Apple, unless of course they acquired the IP they need via some other acquisition they have made. Another possible scenario is Apple has determined that ARMs graphics solutions, named Mali, is now good enough they will start customizing their GPU solution using the Mali IP under the ARM architectural license they already have with ARM.

What many people don’t realize about Apple’s solution with Imagination’s GPU IP is they have been doing a great deal of customization of the Imagination GPU IP to fit their own needs. Apple does not use the generic, off-the-shelf GPU solutions from Imagination but instead customizes it heavily. The work they do to customize their GPU around the Imagination IP is not that different than the work they do to develop their own custom ARM solution using the ARM architectural license.

The folks at Anandtech empahsize this point in an article they wrote today about this news:

Previous to this, what little we knew of Apple’s development process was that they were taking a sort of hybrid approach in GPU development, designing GPUs based on Imagination’s core architecture, but increasingly divergent/customized from Imagination’s own designs. The resulting GPUs weren’t just stock Imagination designs – and this is why we’ve stopped naming them as such – but to the best of our knowledge, they also weren’t new designs built from the ground up.

I concur with their assesement that Apple had not built new designs from the ground up but had been customizing the designs under the license agreement they had with Imagination.

It is unlikely Apple is starting from scratch. However, it is unclear what base IP they will use for this solution. We won’t know until Apple comes to market with the new designs and even then it may not be clear. But if anyone can make this switch and support it with software development APIs and build a base for the future software and services, it is Apple.

Why Design a GPU?
I briefly want to touch on why this important and the main reason why it was predictable. The GPU is actually more important than the CPU with where we are headed in the future. In fact, an underappreciated observation of the last five years in silicon CPU/GPU architecture design is the increasing amount of space the GPU has been taking on the SoC every year. More and more software is leveraging the GPU and, as our core computing expereinces become more visual, the GPU becomes even more important.

The GPU is the heart of not just graphics but advances in imaging technology, visual computing, recognizing images and pictures, machine learning, and AI, to name a few. The GPU sits at the heart of all the things the industry is getting excited about with the future of technology. The GPU is essential in autonomous driving as well.

Apple is putting themselves in a position to own their GPU solution and, as a result, not be beholden to the design direction of a third party whom they may not have total influence over. This puts Apple in an incredibly strong position strategically to control their own destiny in hardware, software, and services — even more than they do today.

Everyone in the semiconductor industry applauds the quality of Apple’s CPU architecture and they have established themselves among the leaders in CPU design. With the leading edge solutions they have developed today and how that has yielded them a significant hardware and software advantage with only designing the CPU, imagine what they will be able to do when they also fully design the GPU.