Today’s Topics: Apple Silicon and Moore’s Law, The One Bundle

Apple Silicon and Moore’s Law
Who would have thought that Apple is the company likely keeping Moore’s Law alive? Noting Apple’s aggressive adoption of new process technology from TSMC, I made the point years ago that Apple was taking over as the company to bring to market Silicon on cutting edge process technology, which was something the industry always looked for Intel to do. But now it is also clear they have taken another thing from Intel, which is Moore’s Law.

I’m not going to debate whether Moore’s Law is dead here or not, but in isolation, if we just looked at Intel, we would have to conclude it had dramatically slowed. In the 20 years I have watched and studied Intel as an analyst, they had constantly pressed forward with Moore’s Law and staying on a roughly two-year cadence of moving from one process node to the next. They called this their tick-tock cycle. Tick was the year they moved to the next process node, and tock was the year they optimized the process node with their architecture. This was up until an array of challenges faced Intel, and Tick Tock became Tick tock tock tock.

But now Apple has been transparent enough with both their process node transitions with the A-series processors, and the number of transistors they are putting on each node, to be confident Apple has been on a tick-tock cycle with their silicon development and A series architecture. At least for the last cycle. The A11 and below did not follow a roughly doubling of transistors. But with the jump from 7nm to 5nm, we see the transistor count did roughly double.

Apple Silicon:
2018 A12 7nm = 6.9b transistors
2019 A13 7nm = 8.3b transistors
2020 A14 5nm = 11b transistors

Now Moore’s Law is more than just transistor count. It also states the cost per transistor will go down as well roughly every two years. We don’t know the economics of Apple’s silicon costs, so it is hard to know if costs are going down, I suspect not, but the transistor doubling seems evident.

It is important to state that doubling transistors for the sake of doubling transistors is not helpful. What matters, and why it matters that you get more transistors in a process transition, is because of how many more operations per second you can run when you have more computing available to you. This is why how a company spends its transistor budget, how they put their transistors to work is more important.

This is where Apple’s silicon architectures come into play. And that is why they show a chart like this every time they release a new A-Series processor.

Apple uses a chart like this to highlight the unique customizations they are putting into their architecture. This chart is a rough highlight of how they spend their transistor budget. Over the years, the number of specific customizations Apple has made has grown from around a dozen to roughly two-dozen. This is insightful as it shows us that as Apple has more transistors available to them, the more customizations they can make for their needs. I pulled this quote from yesterday’s event when the details of the A14 were discussed:

Our goal is to build chips with industry-leading performance, powerful custom technologies, and extremely efficient use of energy to make every one of our products best in class.

The above statement is the foundation of how Apple will spend their transistor budget, and the last line here is truly critical – to make every one of our products best in class. This is why Apple Silicon is so critical. It is the very reason their products will remain best in class.

Hopefully, I can tie all this together succinctly. When Intel, AMD, Nvidia, Qualcomm, etc., all these companies design their architecture and spend their transistor budget in ways that are designed to cover as many possible bases as possible. These companies need to build general-purpose products that can cover a range of customers’ needs, and they do a tremendous job at this. But this comes at a cost. Building a general-purpose architecture means just that it is a general-purpose. Apple built a specific purpose architecture with only one customer in mind, Apple. This allows them to make fundamentally different decisions on how they spend their transistor budget, and it is ultimately the fundamental reason Apple products will outperform other products in the key areas Apple desires and ultimately the reason they can confidently say their products are best in class.

The One Bundle
Apple released the long-awaited bundle of services. From day one, many of us knew this was inevitable, and a number of studies we have done on Apple services made it clear a good portion of Apple services users wanted a bundle.

Apple’s One Bundle pricing is fascinating to analyze. What they included and what prices are quite telling. But the tried and true bundle pricing psychology is well implemented here as Apple has one package that is designed to come across as the easy choice, which is Premium. Here is a chart I snagged from a Morgan Stanely report late last night.

I’d wager a bet, and family plans are quite successful for Apple. Even if you are single, you are likely to be on a family plan with other single friends, or family, etc. I’m sure Apple knows this, hence the attractive pricing for families across the board. The interesting factor here is storage. iCloud storage may be the most attractive hook here, and it lends itself toward Premium as well. I’ve seen research studies suggesting high adoption of iCloud storage but at the .99c and $2.99. I’ve also seen sentiment studies that suggest Apple customers feeling the pain of running out of storage but finding $9.99 or higher for multiple TB of storage harder to justify. The higher tier of iCloud storage makes the premier package even that more enticing.

Assuming I’m right, and more customers choose Premier, it is also strategically brilliant by Apple. I’d again bet, most people who choose Premier have not tried half the services included. They are likely an Apple Music subscriber (a number I think is well north of 100m by now) and an iCloud storage subscriber. Which means they “could” feel included to try these other services, which I think are pretty good, and end up getting hooked. This, for the overall Apple ecosystem, is a good thing for Apple and its grand ecosystem of hardware, software, and services strategy.