Macs are transitioning from Intel to Apple Silicon. Why? And why now?

Are Intel-based Macs still worth buying?

Image for post
Image for post

Apple announced at WWDC 2020 that it will make yet another big transition on the Mac: Away from Intel processors, welcoming Apple Silicon. A little more technical: Away from x86_64, welcoming ARM64. As an effusive user of Apple products, some friends and colleagues asked me some questions about this topic:

What do you think about this change? Is it for the better? What could be the impact on future Macs?

Is it still worth it to buy Macs with Intel processors? Is the price dropping? How long will Intel-based Macs be supported?

The short answers (TL;DR)

I’m very excited, this change will be very positive for the entire Mac lineup. The imminent impact on the next Macs will probably be: far superior performance, better battery life, FaceID on the Mac for the first time.

Yes, Intel-based Macs are still worth buying, because the prices aren’t likely to drop, and official support for x86_64 should range from 2025 to 2030.

The long answers

From a software standpoint, this is going to be very interesting. Every time there is a platform deprecation, it’s also an opportunity to rethink dependencies. The libraries that add universal binary support sooner will earn more users, and the older unmaintained ones will slowly be left behind. With the tools we have today, supporting two platforms shouldn’t be too complex.

Apps should share more code between platforms, given that they are going to be compiled to the same architecture across the board. Apple is already showing that a project’s inside Xcode is going to be organized differently: you shouldn’t think about an iPhone app anymore. You’re going to think about an “Apple Platform” app, and it’s going to have a face on the iPhone, another face on the iPad, and yet another on the Mac. The interface and interactions are going to be different, but a lot of business rules are going to be the same. All code in the same project in Xcode, in one repository.

This is going to increase the number of Mac apps in the long run, because converting and iPad or iPhone app to the Mac is a question of adapting the user interface, leaving the shared business rules practically unchanged.

In order to understand the changes from the hardware standpoint, we have to take a trip down Apple’s culture and history.

Apple has the very strong culture of releasing technology only when it’s ready. Not before, because it wouldn’t have enough quality. Not after, because competition would have already come too close. And once the technology is ready, they launch it as quickly as possible throughout all product lines. This is easily proven by some recent launches:

In 2015 Apple launched the Taptic Engine, which is a technology that makes some iOS controls gently vibrate in a very precise form, so that one can even identify which app is notifying solely by sensing the vibration. In the same year, Apple launched the 12" Macbook that had a new TrackPad taking advantage of the same Taptic Engine technology to trick your brain when you clicked on it: there was no button but your finger felt like there had been a physical click. This same technology has been integrated, in the same year, into the Apple Watch that gently vibrated on the wrist: one type of vibration for email notifications, another for messages, another for alarms, yet another for fitness apps, etc.

In 2016 the True Tone technology was launched on the iPad Pro: it makes the display’s color calibration change in cues of yellow, blue and white, according to ambient light, so that reading text is much more natural to the human eye which expects reflections on surfaces to be in the same color as ambient lighting. In the same year already it was launched on the MacBook Pro. It arrived on the iPhones in the next year.

In 2013 Apple launched iPhone 5S and with it came TouchID. Right after, in 2014, it arrived on the iPad Air 2 and it featured in every iPad model after it. But on the Mac it arrived only in 2016… Why did it take so long?

For TouchID to work securely, there were needed hardware changes to accommodate a Secure Enclave embedded in iPhone’s System-on-a-Chip (SoC). These chips aren’t used on the Mac, that sports only Intel processors. So Apple had to develop a new separate chip, called T1 (and then afterwards T2), to store a Secure Enclave directly on the motherboard, given that they couldn’t alter Intel chips.

This made the fingerprint authentication technology, that was production-ready in 2013 and mature in 2014, be incorporated on the Mac only in 2016. Intel has been hindering innovation on the Mac since at least those times.

In 2017, keeping up with new advancements in technology, Apple launched FaceID, a facial recognition technology that uses many special sensors and a very good front camera. Additionally it depends on a Neural Engine, a brand-new area of Apple’s updated SoC, optimized for Machine Learning.

Right after, in 2018, iPads got FaceID. But for Macs, until today (2020), there’e no sign of the technology, because Intel chips don’t support anything like the Neural Engine. Also that’s probably why 2020 MacBooks still have jurassic 720p FaceTime cameras.

If Apple should have already abandoned Intel long ago, why is this happening now? Many reasons must be involved, of course, but one of the main ones I think is the Thunderbolt license.

Thunderbolt technology is a big competitive differentiator to professionals that work with big files, like audio, video and 3D modelling. Thunderbolt even created a niche in the peripherals business for storage units supporting the technology. For instance, a Thunderbolt 3 compatible device allows transfers up to 40Gbps in a single connection, or even support two 4K displays with a single cable, which is impressive. But it’s Intel’s proprietary technology.

No iPad or iPhone supports Thunderbolt, because Apple’s SoC isn’t compatible with it. In order for Apple to move the Mac to ARM without jeopardizing Pro users that had already purchased Thunderbolt-enabled peripherals, and are used to the high transfer speeds, Apple would have to implement Thunderbolt technology in their own chips, which isn’t really worth it because, apart from having to pay royalties, Intel hasn’t been able to establish Thunderbolt as a market standard.

This makes so much sense that in 2018 Intel stopped charging royalties on Thunderbolt, in a desperate attempt to increase technology adoption, and in 2019, after realizing that the situation was unsustainable, allowed the protocol to be added in the USB-4 open standard.

What a coincidence! Right after that, in 2020, Apple announces the transition to Apple Silicon, so that now they can take advantage of Thunderbolt’s performance by improving their SoC with support for an open standard, royalty-free, with the added side-effect of making sure professional peripheral makers have all the incentive to support it in their new products.

With their own chips, Apple can make huge leaps in the Mac’s computational power. If recently the new iPad Pro has surpassed the MacBook Pro in performance benchmarks, and the iPad doesn’t even have a cooling fan, it becomes easy to project that Apple could yield improvements of at least 30% above the iPad Pro in a MacBook with a more favorable thermal architecture.

Beyond performance, controlling the SoC allows Apple to include high efficiency cores, that consume much less power, to handle more basic tasks, just like the ones already existent in iPhones and iPads. This new Mac could last a lot longer in a single charge when performing basic everyday tasks, but as soon as a heavier application requested more performance, it’s able to deliver just as quickly.

The same could be said about the integrated GPU, that in the iPad is already incredible. I would go as far as to say that dedicated graphic cards on the Macs are going to be reserved to the top-end configurations like iMac Pro and Mac Pro. Additionally, they can add Neural Engine in Mac’s SoC to make very complex tasks, like image and video editing using Machine Learning, be executed almost instantly, perhaps even in real time, just like some apps already do on the iPad.

Finally, it will be possible and viable to bring FaceID to the Mac, opening the way for new technologies, that are yet to be released, to roll out to the Mac at the same time as other product lines.

Price-wise, by my analysis Macs aren’t going to get cheaper. On the contrary, prices are likely to go up. New Macs are going to offer much more features: More performance, more battery life, FaceID, etc.

The appeal to upgrade won’t be because the older Macs don’t work well and are obsolete, but because the new Macs are that much better. And upgrades like this mean price hikes. It’s a similar movement to what they’ve made to iPhone X: they launched iPhone 8 at the same price point as the iPhone 7, and created a new premium price point for iPhone X.

Apple announced that they still have Intel-based Macs on their roadmap, and I bet those are going to stay at their current price points, so that the new Macs with Apple Silicon and new hardware and software features are going to create a new premium price point. Over time prices can go down, specially near the end of support for Intel Macs, so that Apple can drive those older Macs out of the market. But at start, I don’t think so.

Let’s analyze the architectural transitions Apple has made until now:

In the PowerPC to Intel transition, that started in 2005 and went until 2009, Apple supported PowerPC on MacOS for 4 years after the introduction of the first Intel Mac. Back then the number of Macs running on the market was much lower than today.

Apple released 64bit support on iOS in 2013 with the iPhone 5S and stopped supporting 32bit in 2017. Again a 4-year support, on iOS which is a very well-controlled platform with only a few devices available.

On the Mac, 32-bit to 64-bit transition was very different, it took 10 years. This situation is more similar to what we are going to see now, because it’s relatively easy for developers to support both architectures without much hassle.

The exact timing for Intel Macs official support is going to depend a lot on how well the new Macs are going to sell. Apple is basically forced to support Intel Macs while they are the majority on the market. History tells us that a minimum of 4 years can be expected, so support should be official until at least 2025. But the new Mac Pro released last year comes with an Intel chip and it seems to have been built to last a decade. Hence I think it’s more likely that the support window extends until approximately 2030, rounding up to 10 years, similar to the x86 to x86_64 transition.

I’m very excited, this change will be very positive for the entire Mac lineup. The imminent impact on the next Macs will probably be: far superior performance, better battery life, FaceID on the Mac for the first time.

Yes, Intel-based Macs are still worth buying, because the prices aren’t likely to drop, and official support for x86_64 should range from 2025 to 2030.

Written by

Software Craftsman — Engineering Lead @ Mercos

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store