To the casual observer, the excitement around Apple Silicon is concentrated around the specific benefits to Apple and the competitive advantage awarded to Apple thanks to their efforts in custom silicon. There are, however, much wider industry implications.
I have been extremely bullish on Apple’s custom silicon efforts since before Apple silicon began showing up in iPhones. My analysis at Apple’s time acquiring PA Semiconductor stated the writing was on the wall for deeper vertical integration by Apple into more of their products. I vividly recall being at Intel not weeks after Apple’s acquisition of PA Semiconductor. The consensus was quickly established that Apple would make custom silicon a cornerstone of their differentiation and competitive advantage.
Since Apple has now proven they are a force to be reckoned with when it comes to custom silicon, a trend I’ve long predicted is about to accelerate.
Competing in the Race Requires Components Built In-House
A great analogy came to me as I was watching AMD’s CES keynote. AMD CEO Lisa Su invited to the stage, virtually, Formula One race car driver Lewis Hamilton. Hamilton’s presence on stage was mostly to talk about his love of gaming and how AMD helps push the bar in PC and console gaming. While trying to affirm AMD as an innovative technology company, he also made a fascinating point by remarking on how his own industry consists of highly customized components. He stated that all Formula One cars, which extend to all professional race cars, consist largely of components all built in house. The obvious reason for this is a competitive advantage. Every race car driver wants an edge above the competition, and the industry itself has established the only way to do that is with highly customized parts, not stock ones.
To establish this analogy more clearly to the tech industry, Apple products are more like Formula One race cars, which compete against a field of competitors using off the shelf components. As Apple goes deeper into in-house parts for their products, it will inevitably cause others to follow whether they compete with Apple or not.
This is the trend that will accelerate that I mentioned above. While it will be harder for most tech companies to go as deep as Apple in verticalization, I anticipate a much more widespread move to customization.
In an article last year, I dug into this broad semiconductor trend where I highlighted the importance of specific purpose chips over general-purpose ones. The computer industry was built largely on general-purpose silicon. The increased need to differentiate and compete on different planes drives the need away from general-purpose silicon and more to specific purpose chips that are used uniquely by product companies.
I made the point that I don’t think other tech companies will go as deep as Apple in semiconductor customization. Still, I think companies will be at a strategic disadvantage if they don’t include some custom or specialized chips that enable some unique product advantages.
Examples of this trend starting to accelerate have been popping up more often. Microsoft has worked closely with Qualcomm and AMD to offer specific customization to the chipsets that run in certain Surface products. The goal of that work was to differentiate these devices from the pack. Microsoft has even been rumored to be looking more deeply at designing, or co-designing, their own Arm processors, which could have both server and client impact. For many generations of Xbox, Microsoft has created some custom components and co-processors for years as well.
Amazon and Google have become more aggressive in custom/specialized processors to differentiate their cloud platforms. Amazon with its Gravitron processors, and Google with Tensorflow. Both key initiatives to help them compete more uniquely with specialized components and set themselves apart from competitors who use generic parts.
Qualcomm made some news recently that emphasizes this trend when they acquired Nuvia last week. Ex-Apple silicon engineers founded Nuvia, and they created a custom Arm architecture they intended to use for the server market. With their Kryo architecture, Qualcomm used a semi-custom design from Arm but mostly using a more generic Arm IP. Qualcomm recognized their ability to compete depends on more deep customization of technology at every SoC level. This Nuvia acquisition puts them back on the path to more deeply customized semiconductor offerings.
Years ago, the belief that more custom or specialized components were not needed because all computing would be in the cloud someday had never manifested and proven to be a much far off future even if it happens. What has become exceedingly clear is that differentiation at the software level is no longer enough for the tech sector’s biggest competitors. The idea of a “full-stack” company must now extend into components as well.
The race car analogy could not be more apt for the tech industry as the competitive environment increases in every market. The simple truth is companies will need to invest more in specialized components because the nature of competition will demand it.
The bottom line is the more companies create “in-house” technology that extends down into components, the better their chances at competing. The more companies rely on generic solutions, the more they run the risk of falling behind.