Apple (NASDAQ:AAPL) has disrupted several industries over the last decade or so. Most recently, the introduction of the iPad has launched the meme "the death of the PC." If the iPad was right hook to the gut, knocking the wind out of the PC, will a new "MacBook Ether" be the knockout punch?
In my previous post, Apple Developing New MacBook Ether?, I noted that web site Quartz suggested that Apple might ditch Intel (NASDAQ:INTC) chips in its MacBook Air line in favor of its own A-series chips.
I responded that I think Apple would more likely preserve the Air series as a low end "full-powered laptop," and create a new line of lower priced "internet appliance" type computer. I have fancifully dubbed this hypothetical product the "MacBook Ether."
I repeat from my article:
The MacBook Air, first released in 2008, ushered in the era of ultra light weight laptops. It began as an expensive toy for the executive and airline warrior, but as the Price of SSDs (memory based "hard" drives) came down, the Air now represents the entry level laptops in Apple's Mac lineup. Apple's Air pretty much defined the Ultrabook category of computers, several years before the name "Ultrabook" was used.
Today, it dominates the "thin, light laptop" product space with a whopping 56% market share. [source]
Another popular form factor, a subset of the "thin, light" category, is the Chromebook, a set of laptops running Google's (NASDAQ:GOOG) Chrome operating system, and designed to act as an "Internet appliance." That is, they are used mostly to interface with the internet for browsing and email, and to run cloud based applications, specifically Google Docs, but any other web based apps will do. While the new Samsung (OTC:SSNHY) Chromebook Pixel is a powerful machine and costs over $1400, most are low priced devices, such as Acer C7 C710-2847 Chromebook which sells on Amazon (NASDAQ:AMZN) for $252.
Some Chromebooks use low end Intel CPUs, other use ARM Holdings (NASDAQ:ARMH) designs system chips which are used in a majority of smartphones and tablets. And herein lies the crux of the issue.
A brief explanation: ARM Holdings does not make chips. They have intellectual property of a set of designs for chips. If you want to use a CPU of their design, you must either buy one from someone who does make them, or make them yourself and have a foundry produce them for you. Existing chips from ARM designs are the Snapdragon line from Qualcomm (NASDAQ:QCOM), and the Tegra line from NVIDIA (NASDAQ:NVDA). Samsung also makes ARM based chips both of their own designs as well as Apple's own design - the A-series line.
For smartphones and tablets, the ARM CPUs are typically not produced all by themselves. They are produced in a single, complete package with many other necessary components in what is called a "system on a chip" - or SoC. This contains virtually the whole computer portion of the device in one package, the CPU (central processing unit), GPU (graphic processing unit), RAM memory, communications controllers, etc. etc. (That is to say, parts such as a camera, telephone radio components, motion detectors, etc. will still be separate units.)
Apple has been producing their own ARM based SoCs for several years now. Their first in the A-series line, the A4, was introduced with the iPad in early 2010, and the iPhone 4 later that year. The most recent release is the A6X used in the latest iPad. Apple surprised most people when for their latest chips, the A6 and A6X, they moved to their own in house core designs. They licensed only the instruction set from ARM, and redesigned the layout of the chip. This allowed them to build, at a very fine grained level, a chip that was tuned to exactly their specifications. (AnandTech has great reviews here and here.)
It should be noted that Apple acquired P A Semi in 2008, and Intrinsity two years later. Both were small, high performance semiconductor design houses. More recently, MacRumors reported just weeks ago that they have "hired at least a dozen former AMD graphics engineers." Thus, they clearly are aiming at beefing up their design division. The new graphics hires for a special Orlando, Florida team is particularly interesting.
Apple is happy with their Intel based products. Currently, all their full scale laptop and desktop computers run on Intel chips, and they have a cooperation on the ultrahigh speed Thunderbolt cable interface system, that far outperforms USB.
The difference here is cost. For the current MB Air, Intel's Core i5-4250U lists for $345. While I am sure that Apple does not pay that much, is still more than the ARM based chips. Research firm iHS iSuppli estimated that the A6 processor in the iPhone 5 costs only $17.50.
By moving to its own processor, Apple could easily cut the price of an entry level model by at least $300. Other savings could cut it even further. (For complete financial analysis, see the follow up article New MacBook Ether - #3 Financial Implications.)
4- Design - what would it look like?
I think that the new MB Ether would look quite a bit like the Air. Apple has two conflicting issues here. They want to maintain their high end brand image, yet differentiate strongly from the Air.
For the above reason, the Ether would continue the aluminum frame, but would likely add some colored highlights. Ostensibly to add style, the main purpose here is so that people can easily determine that the model is not the more premium Air. This would help to cut down cannibalism.
The new MB Ether would include a new entry to the A-series chop line, the A7XX. Every year Apple has more or less doubled the power of its processors. The A7 group will do that once again, by bringing a quad core to the processor.
The 'X' denomination of the A5X and A6X processors signify that they have extended graphics capabilities to handle the larger screen area of the tablets that they serve by adding more cores to the GPU. Similarly, I envision an 'XX' notation to signify further increased graphics to service not only a larger screen, but the capacity to handle an add on external monitor via a Thunderbolt interface. This capability would distinguish them from the competition.
The big question is what operating system would it use? The first thing one would think is that it would be iOS - what runs the iPhone and iPads on other ARM based systems. The big problem here is that iOS is a touch based system. This would not be the case for the laptop Ether. Therefore, the system would be standard Mac OSX. Perhaps this is one reason that the new version Mavericks is so oriented towards performance optimization. This leads to not only quicker running programs, but significantly better battery usage.
Some have noted that OSX does not run on the ARM based architecture. But this ignores the fact that both operating systems are based on the same BSD Unix cores. Therefore it would not be terrible difficult to port OSX to the ARM processors. In fact, I would be rather surprised if they have not done this a couple of years ago.
Any computer that can access the internet can access any of the browser based storage and apps-in-the-cloud resources. These include storage such as Dropbox, Amazon Cloud Drive, Microsoft's (NASDAQ:MSFT) Skydrive, and also applications such as Google Docs, and Microsoft Office 365. These provide user applications that reside on the internet for editing text documents, spreadsheets, and presentations. Each app - typically written in HTML5 - is run from a compatible browser. The user need not install the application on his personal computer. This saves the user disk space, and harnesses the compute power of the hosting server. Windows 8 provides integrated access to their Skydrive and Office 365, while Chromebooks favor Google Docs. (Neither operating system excludes the use of the other's cloud services.)
Apple, however, just announce at the last Worldwide Developers Conference (WWDC) that their own productivity suite, iWork, will now be available in a cloud version. While currently the three apps (Pages, Numbers, and Keynote) are available only to registered developers, a demonstration was given at the conference keynote presentation that included access via a Windows 8 PC.
The presentation was absolutely striking. Whereas Google Docs and Office 365 look very staid, the iWork interface is clean, clear and open. Its presentation is vibrant. Whether working on a document, a spreadsheet or a presentation, you can just drag an image file off your desktop and drop it into the open document. The system downloads it and places it smartly into place, performing complex word wrap if required. MS Office documents can also be opened by simply dragging and dropping into the browser. It is slated to be available in the fall of this year.
Clearly, if Apple really is ready to market an MB Ether, then they will soon be ready to host their own users with style.
5- Contrary opinions
There have been two major push backs on this idea. First, that Apple does not want to enter the low end market, for reasons both of image and economics. I will address both of these in the financial analysis of the next article.
The other major objection is that the whole thing is impossible because this whole hypothesis is based on the idea that Apple will be buying new A-series chips from TSMC, and these can compete with new offerings from Intel. As noted in the previous article:
Taiwan Semiconductor Manufacturing Company (TSMC) and its IC design service partner Global UniChip have secured a three-year agreement with Apple to supply foundry services for the next A-series chips built using 20nm, 16nm and 10nm process nodes, according to industry sources.
This was further elaborated by the Wall Street Journal.
Yet some have replied to my initial post that this cannot be true since TSMC will not be able to move to the designated processes in time. (Note: the nomenclature 20nm or 16nm refers to the size of transistors etched into the silicon on a chip. Obviously, the smaller the transistors the more you can get in a given space, or the smaller you can make a given design. Additionally, typically a device on a smaller technology requires consumes less power.)
Arnold Frisch (former Tektronix and Intel engineer) writes that TSMC, GlobalFoundries, And Samsung Can't Save ARM And AMD From Intel's FinFETs. (FinFET is a particular type of chip technology.) According to this, ARM technology has no future.
This is echoed by Intel chief of mobile Mike Bell in a very interesting post by Sebastian Anthony of Extreme Tech. Bell tells the author:
"I see no data that supports the claims that ARM is more efficient."
…I finally ask the question I've been gagging to ask: Where does this leave ARM, ARM's licensees… and the foundries? If Medfield is already competitive on power consumption and processing power, and Intel has such a dominating technological lead, where does ARM go from here? "I think," Bell begins slowly, picking his words carefully, "Moving forward, it will be difficult for anyone who doesn't have an end-to-end capability to keep up with us. I took it for granted before I joined Intel, but this really is rocket science. When you see people working on 9nm - I see the guys in their bunny suits, doing the mask generation for the chips - you realize this is probably one of the most difficult industries I've ever seen. There are very few companies on Earth who have the capabilities we've talked about, and going forward I don't think anyone will be able to match us." [emphasis added]
In short, Bell says that there is no future for ARM technology.
Yet, if Digitimes and WSJ are correct, then Apple (not known for being blind fools) has signed up TSMC to do precisely 20nm and 16nm FinFET production. Jean-Louis Gassée (former Apple exec) writes for the Guardian as to why Intel has not jumped on a chance to manufacture Apple A-chips.
The first explanation is architectural disdain. Intel sees "no future for ARM", it's a culture of x86 true believers.
And they have a right to their conviction: With each iteration of its manufacturing technology, Intel has full control over how to improve its processors. They can reduce x86 power consumption by using smaller building blocks (they're already down to 22 nanometers wide). They can micro-manage (literally) which parts of a complex chip will be turned on, off, or somewhere in between - in a kind of hibernation.
So, while he does say that they may have a point, it does not completely free the argument from heavy personal bias of the "x86 true believers." Even if, as Bell states, it is rocket science, there really are more "rocket scientists" than those who work for Intel.
Gassé points out:
- As evidenced by the A6 processor running inside the iPhone 5, Apple goes to great lengths to customise the basic ARM cores, adding graphic processors, memory, and large amounts of support logic, and even resorts to aggressive hand-optimisation [sic] of the silicon layout - as opposed to just letting CAD software tools do the job.
He feels that this is a bad mix with Intel's culture, and hence the lack of partnership here.
But this is an aside. My point here is just that this is something that remains to be seen - whether or not ARM production techniques can keep up with Intel's.
Apple certainly is able to build serious new laptops based on it ARM-based A-series CPUs that will provide a quality experience. Doing so, they will finally enter the low cost computing world, opening up a whole new market.
They have the technology. They have the overall ecosystem - particularly with the new iWork in the cloud suite. So two questions remain:
- Do they have the desire to do so?
- What are the economics?
The first is something few outside of Apple (or even inside) can know. The economics question is what I address in the third (and final) post in this series.
It should be noted here that there are two ways that a customer can purchase ARM designs. They can purchase complete chip designs, more or less ready for a fab facility to produce or to be used as a portion of a more complex package; or they purchase only the instruction set and then design their own silicon to implement the instruction set. Most ARM based chips do the former, use complete core designs. Apple breaks this mold with the latest A6 chip and created their own, original core design according to their specific priorities.