Apple Silicon: FAQs and Future

Apple is starting to move its Mac line away from Intel chips to its own silicon. Here are answers to some of the questions I have gotten from clients and press about the move, the new Macs, and how it impacts competitors and the market going forward.

What are Apple’s goals for the M1?

While Apple will undoubtedly save money in the long run by designing its own SoCs, Apple is making the switch for control and competitive differentiation, not cost savings. The move to bring the Mac’s processing in house was inevitable, but Intel fumbling process improvements over the past five years has sped up the transition by several years. Apple Silicon gives the company more flexibility on price, but there are still R&D costs to recoup, and, in any case, Apple is determined to keep Mac pricing solidly at the premium level. Apple already has a lower cost computing platform, it’s called “iPad.”

On a technical level, Apple is balancing speed and battery life by bringing as much as possible into the SoC package, including integrated system and graphics memory, security, AI, video processing, IO, and more. It’s a lot to cram into one SoC, and Apple is using TSMC’s very latest 5nm process, which keeps the physical size down while speeds and efficiency are up. Apple also appears to have specifically engineered the M1 to speed up fundamental MacOS tasks – a good example of the integration between hardware and software that Apple is always touting. The initial round of reviews and benchmarks for M1-based systems suggests that Apple has succeeded in thoroughly trouncing Intel processors designed for thin/light laptops while adding hours and hours of additional real-world battery life. Apple claims that the M1 has the highest performance-per-watt of any SoC in the world, and that claim appears to be credible.

Despite some Amazon-style graphs with few details attached, Apple is not attempting to out-perform Intel and AMD’s highest-end chips and systems with discrete graphics with the M1. While reviewers found that the M1 can seamlessly support workloads with endless browser tabs, managing enormous image or video files could hit the hard limit of 16GB of integrated RAM. Apple will have a separate class of processor designed for those use cases in mobile products, and could field yet another variant aimed at high performance desktops with fixed power.

Apple is also not integrating cellular connectivity on the M1, though there is nothing preventing Apple from doing so in the future. (Apple bought Intel’s modem business, which has LTE capabilities today, but is several years behind Qualcomm on 5G.)

Is Apple Silicon a transformative industry event?

Yes and no. PCs are a mature market, and even if Apple massively increases its market share in laptops and desktops, that won’t have much impact on the computing market overall. Qualcomm and Microsoft were already pushing Arm further into laptops, and Nvidia badly wants to get its Arm designs into the data center. Apple isn’t selling M1’s to anyone else, and while it is likely to pull in a lot of switchers from Microsoft, Windows is not the center of Redmond’s business anymore. The $999 MacBook Air was already Apple’s best-selling laptop, and sales are sure to grow in education and the enterprise. However, the bulk of PC sales are well below the MacBook Air’s price points, and there is a limit to how much of the market Apple can convert to premium. Most buyers are still faced with the binary choice of MacOS or Windows, and only then is the silicon architecture a consideration. There is an outside chance that Apple will use the M1 to attack budget laptops, but, again, that’s what the iPad is for.

Still, Apple Silicon appears to be amazing. Apple’s Rosetta 2 emulation layer is working remarkably well for early reviewers, and Apple is likely to be successful getting key third party apps rewritten to work natively on Apple Silicon within the next 6 – 18 months. The Mac moving to Apple Silicon will have an outsized impact on several companies in the PC and semiconductor industries, and Apple’s successful example will push Arm architecture into PCs, data centers, and edge computing faster than it would have otherwise gotten there.

Why are the new Macs so boring?

Apple is completely changing its software and silicon architecture, so it deliberately left the hardware form factors alone to minimize further risk. Once the transition to Apple Silicon has been successfully made, I expect Apple’s designers to rethink what a laptop with instant-on, endless battery life, and significantly cooler thermals should look like. For now, it is sticking with designs that are already known to work.

However, it is instructive that Apple redesigned the user interface to be more like iOS and iPadOS yet did not include provisions for a touchscreen. In fact, Apple created a whole series of trackpad gestures to (awkwardly) control touch-oriented apps. Apple continues to believe that Macs – even Macs that can run iPhone and iPad apps – should abstract the control point from what is being displayed. If you want to directly manipulate the items onscreen, the iPad is the computer for you.

What does this mean for Intel?

It is really bad, but not because Intel is losing sales directly. Apple makes up enough of Intel’s revenue to be noticeable, but not enough to anguish over. Apple is also a notoriously demanding customer and it would not surprise me if its profit margins servicing the business were low. Intel also won’t lose the entire account at once; over time Apple will put its own silicon in all of its Macs, but the Mac Pro will likely be last to fully switch away from Intel. This has as much to do with the software and customer requirements of the Mac Pro as to the chip architecture.

No, it’s bad because Intel was already under attack from a resurgent AMD, and now Apple has completely reset the bar across the industry for performance per watt. Even if Intel does get its process issues back on track, it will somehow need to make enormous jumps just to catch up to where Apple is today. And Apple is not going to stand still. It will get really interesting when Apple starts specifically optimizing its architecture for higher performance in a bigger thermal envelope and constant power for desktops. The improvements on the GPU side will play a big role here, too – how long will Apple stick with AMD, and how competitive will Apple discrete GPUs be?

Intel’s worst nightmare, however, is unlikely to come to pass. Consumers know Intel for its PC chips, but Intel’s growth lies in large datacenters, telecom, and edge applications. While Apple could undoubtedly attack those markets with Apple Silicon, it is unlikely to do so, as that would take the company away from its goal of building tools at the intersection of technology and liberal arts. Apple famously says, “no” to a lot of opportunities, and I believe that Apple would much prefer to spend its management, engineering, and sales time focusing on new computing form factors, consumer services, and AR glasses rather than, say, OpenRAN.

What does this mean for Qualcomm?

On the one hand, Apple is handing Qualcomm a huge gift: Apple has delivered on Qualcomm’s AOPC (Always On PC) promise, proving that the Arm instruction set really can outperform x86 and provide longer battery life at the same time. This undoubtedly will fuel renewed interest in Windows on Arm, and Qualcomm is the leading candidate to capitalize on it. However, Windows on Arm still has performance and software compatibility issues to iron out before it can displace Intel in most PCs. Microsoft's upcoming x64 emulation layer in Windows should help, but Windows also has a lot more enterprise app baggage to deal with. This legacy library is a strength of the platform, but it makes transitions harder. Apple can tell its developers to rewrite their apps and the biggest names will comply, and the compatibility problem is solved. In the meantime, the M1 is so fast that even in emulation it should provide good enough performance for most use cases. However, Microsoft has had a hell of time getting developers to recompile for Arm, Qualcomm has not provided enough horsepower to emulate its way through the problem, and the long tail of enterprise and x86-specific utilities will drag the process out regardless.

It is worth noting that not all Arm designs are the same. Apple has a perpetual license to the Arm instruction set, but it does not license or use Arm’s processor designs like Qualcomm, Mediatek, Samsung, and others do. Qualcomm will be pressed to improve its performance and battery life, but until it can do so, it is likely to focus on integrated connectivity, an area where the Macs currently don’t compete.

To discuss the implications of this report on your business, product, or investment strategies, contact Avi at avi@techsponential.com or +1 (201) 677-8284.