Tuesday, August 26, 2014

Why Apple Will Switch To ARM-Based Macs

When Steve Jobs announced the transition from PowerPC to Intel processors in 2005, he revealed something that, in hindsight, seemed obvious to everyone who didn’t anticipate the switch:

There are two major challenges in this transition. The first one is making Mac OS X sing on Intel processors. Now, I have something to tell you today: Mac OS X has been leading a secret double life for the past five years.

[…]

We’ve had teams doing the “just in case” scenario. And our rules have been that our designs for OS X must be processor independent, and that every project must be built for both the PowerPC and Intel processors. And so today, for the first time, I can confirm the rumors that every release of Mac OS X has been compiled for both PowerPC and Intel. This has been going on for the last five years.

There’s not a doubt in my mind that if you substitute Intel for PowerPC and ARM for Intel, what Steve Jobs said then holds 100 percent true today, word for word. Mac OS X designs must be processor independent, every project must be built for both Intel and ARM processors, and each Mac OS X release in the last five years has been compiled for both Intel and ARM.

Somewhere on Apple’s campus, ARM-based Macs are already running OS X.

User Experience Would Improve

In his iPhone 5S review, Anand Shimpi compared the Apple-designed A7 processor with Intel’s fastest tablet chip at the time. He wrote:

For our cross-platform CPU performance tests we turn to the usual collection of Javascript and HTML5 based browser tests. Most of our comparison targets here are smartphones with two exceptions: Intel’s Bay Trail FFRD and Qualcomm’s MSM8974 Snapdragon 800 MDP/T. Both of those platforms are test tablets, leveraging higher TDP silicon in a tablet form factor. The gap between the TDP of Apple’s A7 and those two SoCs isn’t huge, but there is a gap. I only include those platforms as a reference point. As you’re about to see, the work that Apple has put into the A7 makes the iPhone 5S performance competitive with both. In many cases the A7 delivers better performance than one or both of them.

In September of 2013, the world’s preeminent independent processor expert compared Apple’s latest iPhone chip with Intel’s fastest tablet chip and concluded that the two perform similarly — even though the Intel chip draws more power, contains four cores versus the A7’s two, and is produced with a more advanced manufacturing technique. If Apple’s chip design team can create a phone processor that performs on par with Intel’s fastest tablet chip, the company’s “highest priority”, then there’s no reason to believe that the same team at Apple can’t design chips powerful enough for any Mac in the company’s lineup.

Apple has already released a line of A-series chips tailored specifically for iOS devices, and the company is most definitely working on a line of B-series chips tailored specifically for Macs. When that B-series chip — or set of B-series chips that runs in parallel — is ready, Apple will be able to switch to ARM-based Macs without sacrificing user experience. On the contrary, because the company is no doubt designing its line of B-series chips in tandem with Mac OS X, there would be iPhone-like hardware-software optimization, improving user experience.

Apple Would Make More Money Per Mac And Sell More Macs

Going from chip concept to manufactured product can be broken down into two separate and distinct steps. The first is chip design — figuring out what features the processor will have and how it will work. The second is manufacturing — turning a file that exists on a screen into a physical product you can hold in your hand.

Today, Intel designs the chips in Macs and manufactures them, profiting on both of those steps. But if Apple swapped out Intel’s chips for its own ARM-based designs, an external company would profit on only one step of the chip creation process, not both, leading to a decrease in the cost of building a Mac. By my conservative estimate, Apple would be able to drop the price of the base model 11- and 13″ MacBook Airs by $50 and still make more profit per unit on each than it currently does.

This cost savings would apply to the entire Mac lineup. Apple would be able to drop prices across the board and make more money per Mac than it does today — and with lower prices, the company would sell more of them, too.

Apple Would Be Able To Create Better Macs

When Apple announced the iPhone 5S, it explained that all of the fingerprint data associated with Touch ID “is encrypted and stored inside the secure enclave in our new A7 chip” where it’s “locked away from everything else”.

Apple wouldn’t have been able to create Touch ID if the iPhone were powered by an Intel chip instead of an Apple-designed one. There wouldn’t have been a “secure enclave” on the iPhone’s processor to store the fingerprint data, nor would there have been perfect hardware-software integration. Apple was able to implement Touch ID because it designed the A7 chip in tandem with the iPhone 5S’s software and the rest of its hardware.

I’d bet that there are features Apple envisions for the Mac that simply can’t be built while Intel designs the chips inside of them. To implement those ideas, Apple would need to switch the Mac to ARM-based processors, because only then would the company have the ability to design chips customized for specific features. If Apple moved the Mac to ARM-based chips, the company would literally be able to create better products than it can today.

This brings me to something else Steve Jobs said when he announced the transition from PowerPC to Intel. Ultimately, he explained, Apple switched for one simple reason: “We can envision some amazing products we want to build for you, and we don’t know how to build them with the future PowerPC roadmap.”

The same logic applies today. It’s not a stretch to imagine Tim Cook walking out on stage and saying, “We can envision some amazing products we want to build for you, and we don’t know how to build them with Intel’s chips.”

As I first said more than three years ago: ARM-based Macs are definitely coming.

Wednesday, July 16, 2014

The Times They Are A-Changin’

Intel released its second quarter earnings yesterday, but the company put out a far more interesting press release last week, on July 7th.

Intel PR:

Intel Corporation today announced that it has entered into a manufacturing agreement with Panasonic Corporation’s System LSI Business Division. Intel’s custom foundry business will manufacture future Panasonic system-on-chips (SoCs) using Intel’s 14nm low-power manufacturing process.

What’s incredible about this isn’t that Intel is manufacturing another company’s SoC — it’s that Intel is manufacturing another company’s SoC with its most advanced manufacturing process.

Traditionally, Intel has produced its newest chips with its most advanced (i.e. smallest) manufacturing process. Not only does a smaller manufacturing node make the chip cheaper to manufacture, but it also makes the chip more power efficient. Taken together, Intel has traditionally been able to charge a steep premium on its newest chips in no small part because they were made with the company’s most advanced manufacturing process.

For Intel to make another company’s chip with its most advanced manufacturing process is anathema to the way Intel has been running for a while now. At any given time, Intel has a finite capacity to produce processors at its most advanced manufacturing node. By definition, Intel is giving up capacity to produce its own 14nm designs in order to make 14nm Panasonic chips.

And that’s very interesting. Panasonic will no doubt pay a premium for the advanced manufacturing process. But there’s no way that the premium is more than what Intel is forgoing by giving up capacity that could be used to produce 14nm x86 chips. 

With this deal, Intel consciously chose to forgo production of its most advanced and profitable chips in order to make pocket change on Panasonic SoCs. That makes no sense.

But there is another possible explanation.

Years ago, when Intel was designing the factory that would eventually produce 14nm chips, the company decided to make the fab capable of producing X amount of chips in a given period of time based on the assumption that OEMs would purchase a certain number of x86 processors. But things didn’t turn out the way Intel expected them to, and OEMs aren’t buying the number of chips Intel projected they would.

In other words, Intel has excess 14nm production capacity.

Intel, of course, doesn’t want that production capacity to sit there idly. The company would rather produce Panasonic’s SoCs and make a few bucks instead of making no money with that manufacturing capacity at all.

Viewed as an isolated incident, Intel’s excess production capacity isn’t that big of an issue — the company simply incorrectly forecasted demand for its chips.

But this isn’t an isolated incident — rather, it’s a sign of things to come. It’s clear that there will be less demand for Intel’s traditional consumer x86 chips five years from now than there is today. And, like it’s doing now, Intel will make the best use it can of its excess production capacity. Better to produce other companies’ SoCs and make some money with the capacity than let it sit idly and make no money with it at all. 

The problem is that the revenue Intel will bring in by producing other companies’ SoCs won’t be enough to offset the decline in x86 revenue. In fact, the revenue Intel will bring in through fabbing third-party SoCs will almost definitely be less than the profit it would make by selling its x86 chips. Quite simply, I can’t see how the margins on producing other companies’ chips will be anywhere near high enough to support Intel’s 100,000+ employee workforce.

Layoffs are inevitable. All Intel can do at this point is take the right steps to minimize them.

Friday, June 13, 2014

Why x64 Chips Will Never Be In iOS Devices

Jean-Louis Gassée:

ARM derivaties need less circuitry for the same computing task and, as a result, dissipate less power. This is one of the key reasons for their dominance in the battery-powered world of mobile devices. (The other is the customization and integration flexibility provided by the ARM ecosystem.)

If you want to understand why Intel’s x64 chips won’t ever show up in an iOS device, you have to recognize that ARM originally became dominant in mobile for one reason, and that it’s dominant in mobile right now for another reason entirely. I think it’s safe to say that Apple, along with the rest of the industry, originally chose ARM-based chips over Intel’s for JLG’s first reason — their longer battery life.

As time went on, however, and ARM became entrenched in the smartphone and tablet spaces, the architecture’s customizability (JLG’s second reason) became just as important to Apple as its low-power design. It allows the company to create everything in tandem — iPhone hardware, iPhone software, A-series chips. Even if Intel were to show Apple an x64 chip that consumed less power than Apple’s ARM-based chips, Apple wouldn’t switch. Battery life is important, but ARM’s customizability is what allows Apple to be Apple. The most important aspect of a mobile architecture to Apple today is its customizability — and x64 doesn’t offer any.

Monday, April 14, 2014

Apple, Samsung, and Intel

Apple and Samsung have the most interesting relationship of any two companies in the technology industry today, and perhaps the most interesting of any two companies in any industry in recent times. On one hand, they‘re suing the pants off of each other in courts across the world. On the other, Samsung supplies Apple with the component at the heart of the iPhone and the iPad, the processor. This we think you ripped us off and you owe us tons of money but please keep making our processors for us relationship is fundamentally unsustainable. Something is going to happen, and it won’t end well.

I believe it’s in Apple’s best interest to move away from Samsung as the sole supplier of its A-series chips. Adding another supplier (or two) would be better than the current set-up, of course, but Apple should move away from Samsung entirely, and as soon as possible.

Who, then, should manufacture Apple’s A-series ARM-based chips? What other companies even have the capacity to produce the 200+ million processors Apple demands every year, not to mention the millions and millions more per year the company will demand in the future, as it continues to grow? Here is a short but incomplete list of companies that have years of experience producing other companies’ ARM-based system on a chip (SoC) designs: TSMC, UMC, Global Foundries, SMIC, and Tower Semiconductor.

I imagine that few, if any, of those companies could produce enough chips to satisfy Apple’s demand. But there is one company with the ability to meet Apple’s demand, a company with decades of semiconductor fabrication experience, a company that recently started to produce the designs of other companies’ ARM-based SoCs: Intel.

Though most people don’t know it, Intel in 2012 agreed to start manufacturing the ARM-based SoCs of other companies. The company realized its business model was outdated, and, to the credit of then-CEO Paul Otellini, started taking steps to disrupt itself. The company recognized that if it maintained its current business model of selling fully fleshed out x64 chips to OEMs like Dell and HP, it wouldn’t last in the Post-PC era, and that if the company wanted to remain relevant in the personal computer market going forward — and why wouldn’t it? — then it needed to change.

Apple, due to the sheer volume of processors it purchases, will always be at the mercy of its suppliers. But if Apple were to contract the manufacturing of its A-series chips to Intel, no longer would the company be at the mercy of Samsung, a supplier who stands to gain a lot if the iPhone were to suffer from a component shortage. Rather, it would be at the mercy of Intel, a company that has nothing to gain and everything to lose by not producing enough SoCs. That’s a much healthier dynamic for a client-vendor relationship.

Historically, Intel has been a great partner to Apple. It has produced custom chips for the company and has given it short-term exclusives on new processors. Additionally, if Intel didn’t do such an incredible job during the PowerPC-to-Intel transition of 2006, Apple wouldn’t be anywhere near as successful as it is today. The fact that Apple was able to say You can run Windows on your Mac definitely helped convince tons of on-the-fence customers to get one. If it weren’t for Intel, Apple wouldn’t have been able to say that, so tons of Macs never would have been sold, and Apple’s halo effect would not have been as strong as it was, resulting in fewer iPhone and iPad sales. On the flip side, Intel’s worst offense was not finishing the development of its new x64 chips in time for Apple to release new Macs when it had planned to. If Intel were to produce Apple’s SoCs, though, that wouldn’t even be an issue — Apple would be the company designing the chips, Intel simply manufacturing them.

Contrast that with Apple’s relationship with Samsung. Samsung, as a South Korean company, has a lower bar for what’s considered ethical (as measured by Western standards) than its American counterparts. The company has admitted, in court, to ripping off Apple’s intellectual property, and I have no doubt that Tim Cook has considered the possibility that Samsung reverse engineers the processors it produces for Apple before Apple releases them to the public in order to make its own smartphone and tablet offerings more competitive.

It’s obviously in Apple’s best interest to hire Intel to produce its A-series chips instead of Samsung. But is it in Intel’s best interest as well? That’s the question. Unless it’s in both companies’ best interests, a deal won’t be struck.

Intel has already recognized that it is in the company’s long-term best interest to get into the ARM game in one capacity or another. At the same time, Intel no doubt has also recognized that it is not in the company’s short-term best interest to start producing large quantities of ARM-based chips for any company that asks. Intel has only so much production capacity, and the opportunity cost is too high to start producing low-margin ARM-based chips at the expense of its more profitable x64 processors. The company is at a crossroads: its short- and long-term best interests don’t align, and it has to choose one at the expense of the other.

That’s where Apple’s cash hoard comes in.

With its $125+ billion in cash, Apple could build a factory for Intel, changing the company’s calculus. Rather than Intel having to choose between building ARM-based SoCs or x64 chips, it could make both; Intel could manufacture Apple’s A-series processors without giving up any of its capacity to produce its more profitable x64 designs.

This arrangement would benefit both companies in a number of ways. Apple would no longer depend upon Samsung, its biggest competitor, to produce the chips at the heart of its most successful products. (This is analogous to America asking China to build its most advanced missiles and hoping the country won’t use any of the top-secret technology it learns about for its own benefit when it’s clearly in China’s best interest to do so.) And because Intel has manufacturing capabilities that other companies don’t, Apple might well be able to create better chips than it would be able to if it were to continue using Samsung as its chip manufacturer. Finally, the company would have peace of mind knowing that its chip producer doesn’t stand to gain anything from a processor shortfall, as Samsung does. Even if the factory were to cost $5 billion — and it wouldn’t — it’d be worth it. Steve Jobs said Apple’s cash hoard is for “big, bold” “strategic opportunities”. This move exemplifies that thinking.

Intel would benefit from this arrangement as well. The partnership I outlined above — or something similar to it — would allow the company to further disrupt itself without taking a large financial hit. The company could make Apple’s ARM-based SoCs without having to decrease the production of its more profitable x64 chips. Additionally, Intel needs to get into the ARM game in meaningful quantities if it wants to stay relevant in the personal computing market, and from a business-to-business perspective, there is no better way to do that than take on Apple as a client. Intel could tell all of its potential customers We make Apple’s chips, and we can make yours, too. Every other company would want Intel to manufacture their SoCs. Producing Apple’s A-series chips would jumpstart Intel’s new foundry business and give the company an easy path to success in the ARM game, ensuring the company’s relevancy in the personal computing market a decade from now.

Intel today faces a dilemma. It’s in the company’s long-term best interest to be able to say, ten years from now, that it’s a large manufacturer of ARM-based chips, yet it’s not currently in the company’s short-term best interest to start producing them on a large scale. The company would have to stop manufacturing its own high margin x64 chips in favor of the low margin ARM-based SoCs of other companies. In other words, Intel today has to choose between its short- and long-term best interests. But Apple, with its cash hoard, can solve that problem. By building Intel a factory, Apple can give the company a way to pursue its long-term best interest without sacrificing its short-term best interest.

Both companies have so much to gain and practically nothing to lose from a partnership — so much upside, almost zero possible downside. I don’t know exactly when, but I’m convinced it’s going to happen sometime soon.

Tuesday, April 8, 2014 Sunday, March 30, 2014

AYH

Whenever my friend Claire visits me at Temple, we always go out for crêpes. They’re delicious, they’re not heavy, and they’re great with a latte. They might be the perfect food.

While in Washington D.C. this past January, I met up with my friend Chanelle, who I hadn’t seen in years. We went to Baked & Wired, a highly rated coffee- and bakeshop in Georgetown. I ordered a traditional latte and she ordered a chai latte. Being the social eater/drinker that I am, I suggested we swap drinks halfway through; that way we could each taste two different drinks from this highly regarded establishment. She took me up on my offer, and, halfway through, my life changed for the better. I had my first taste of a chai latte, and holy hell was it good — the perfect mix of spicy and sweet.

A month later, my friend Kevin and I went to Elixr, a coffee shop in Philadelphia. Waiting in line, I told him about my chai latte experience. Much to my amazement, he’d never had one before. The solution was obvious: buy him a chai latte.

We got our chai lattes, drank them, and agreed that, while they’re incredible, they wouldn’t go well with everything. Later that day, I realized what they’d go perfectly with: crêpes. Immediately I texted Claire about my discovery. “It has a unique taste. Once you have it you’ll know what I mean. It doesn’t go with everything — though I do think it’d be perfect with a crêpe.”

Eight days later: “I had a chai latte today. It was heaven.” Not only did she like love it, but she agreed it’d go perfectly with a crêpe.

While at the gym the following Wednesday, I felt a craving for a chai latte. And at that point, sweaty and in gym clothes, I knew I couldn’t get one right now like I wanted to. So I did the next best thing: look at pictures of chai lattes on the Internet!

I don’t know how, but eventually I came across this picture:

image

I sent the link to Claire, and though she said she’d really like to drink one, she missed the funniest part of the whole thing: the website I found the image on was deliciouscrepesbistro.com, a crêpe restaurant. And they serve what looks like the best chai latte ever. I pointed this out, and she started laughing. Then she said: “Does it say they are in California? Road trip!”

The thing is, I was already planning a summer trip to California. So when she said that, there was only one logical response: Do you want to come with me?

“That would be EPIC!”

She sent that message at 10:28 AM. And a few minutes later, it hit me: what an absurd idea. What an absurd thing it is that two 18 year olds, who had at that point been in each other’s physical presence all of two or three times, could even contemplate the idea of flying across the country together, to a place where neither of us have ever been. And not just hypothetically — it could really, actually happen. We each have the money; we wouldn’t have to ask our parents for a dime. It’s absurd — absolutely absurd! — and in the best possible way.

When I realized this, I couldn’t help but ask how things got to that point. Why, at 18 years old, could I even consider doing something so ridiculous? There I was, sitting alone with a cup of coffee in my dorm at 10:30 AM on a Wednesday morning, when it hit me: it all comes back to my parents.

While other parents wouldn’t have, my parents encouraged me to pursue my passions. This led to me working in the real world far before most kids my age, which led to me maturing way before my peers. Another benefit: I’d saved up lots of money.

Clearly, it all came back to them. They made choices other parents would not have. And it’s because of those conscious choices that I’m in the position I’m in today. It’s because of those choices that it’s not outside the realm of possibility to fly across the country with one of my best friends.

It’s all because of them.

When I realized this, I started crying. At times I had been a giant dick to my parents, but it’s because of them that my life is so great. I’d fight with them over the stupidest of things, not appreciating how incredible they really were.

Eventually I composed myself, and at 11:04 AM, I sent them this message:

I don’t think I’ve ever said point blank how much I appreciate everything you guys have done for me. The reason my life is so great is because of you guys.

It was you guys who encouraged me to pursue my passions, which put me in the position to learn the skills necessary to succeed. It’s because of you guys that I’ve had the incredible opportunities I’ve had in life. You’ve given me the freedom to live life to the fullest. And when I’ve gotten a little off course, you’ve always nudged me in the right direction.

Despite the shit I give you guys sometimes, I really do love you both. Thank you very, very much for everything.

Two minutes later, at 11:06 AM, my mom responded.

Are you high?????

Tuesday, March 11, 2014 Sunday, January 19, 2014 Wednesday, October 30, 2013 Wednesday, October 9, 2013