Seven Questions for ARM CEO Warren East
It’s kind of hard these days to avoid an ARM chip. There are probably five or more inside your mobile phone alone, a few in your car, some in your PC, and several more in places you wouldn’t think of, like your coffeemaker.
Things are good for ARM Holdings, the British chip company whose designs are central to so many of the chips that make modern life modern. In 2011, some 7.9 billion chips with ARM cores in them were shipped. And yet it’s not a very big company. Where Intel clocked sales of $54 billion, ARM finished the year with sales of $777 million (491.8 million pounds). It all has to do with the differences in how they do business. ARM sells the blueprints to make a core — the central brain of a chip — and then those who buy that blueprint can build their own custom parts of a chip around it.
That means an ARM-based chip from Samsung can be significantly different from an ARM chip from Broadcom or Nvidia. And yet designers from either company could probably exchange jobs, because they’re both familiar with the basic designs. ARM has become something of a lingua franca of electronics design, except in the world of personal computers and servers. Yet with Microsoft set to release a new ARM-friendly version of Windows for notebooks and tablets, and the chip firm Calxeda working on bringing ARM chips to servers, ARM’s influence is growing.
I caught up with ARM CEO Warren East over dinner in New York last week, and we talked about how its business model is going strong, and where the ARM architecture is going.
AllThingsD: When people ask me what ARM is, I tend to liken it to a recipe for cake — a cake for which you buy the basic recipe, but which you can then enhance anyway you like. Is that a fair analogy?
East: Exactly, and the doing whatever you like is very important for our business model. If you couldn’t, and we were like Intel, say, and you had to do this one thing, the only thing our licensees could — if you were to apply a licensing model to that — the only thing they could use to compete against each other is price. Whereas this way, they can do their own stuff around the basic recipe, they can differentiate. But because it’s the same microprocessor architecture, your cake recipe, then investments they make in software, or if you’re using a combination of chips from Samsung and Nivida and Qualcomm, any investment you make toward using Samsung chips is equally applicable to the others.
And you can switch to another vendor later if you like, correct?
You can, because they all do different things. If your product is about video, then Texas Instruments’ video accelerator is very good. If it’s about 3-D graphics, then Nvidia’s chips are very good. If it’s a modem you need, then Qualcomm’s chip is very good. So you can mix and match.
And it’s not uncommon for many manufacturers, whether they’re making phones or something else, to have several ARM-based chips doing many things. In a phone, the main microprocessor will be an ARM-based chip, but then also the surrounding chips doing specialized functions will be ARM chips, as well, correct?
Right. The typical smartphone will have four or five ARM chips in it. There’s the main processor, the thing you interact with as the user. Then there’s the modem, which connects to the phone network. And then there’s a connectivity processor that handles the Bluetooth and the Wi-Fi or both. And then there may be a power management processor, or a touchscreen controller, a camera, or GPS, and so on. And the next one that’s being integrated is NFC, or Near Field Communications, for payments by phone. And your 8-bit processor in the SIM card is turning into a 32-bit microprocessor, and that will likely be an ARM, as well.
When you think about competitors, who is it? Is it MIPS? Is it Intel, perhaps, down the road?
When you think about the consumer electronics space, TVs and the like, MIPS has been very strong in that space. Increasingly, as the TVs become smarter and more connected then they start to look more and more like a smartphone with a 46-inch screen. And so, actually, the infrastructure that exists around ARM makes it very compelling to put an ARM chip in there. In the computing world then, the competition is really Intel and AMD x86 chips.
Speaking of AMD, its CEO, Rory Read, raised some eyebrows at its analyst meeting recently when he mentioned ARM and described a new “ambidextrous” approach to its chips, implying, many think, that AMD might combine its x86 cores in some way with an ARM core. Can you give any visibility into what he might mean?
We can’t tell you really anything about it. But I will say something that we’ve said about this before, when people had picked up similar noises about something like this. AMD is in the business of selling microprocessors. We’re in the business of selling microprocessor designs. We wouldn’t be doing our job properly if we weren’t at least talking to them. And so we have been, for the last 10 years or so. If those discussions go anywhere, and if and when there’s something to announce to the world, we’ll do so.
How many licensees are there? Are there any that surprise you because they’re unusual or unique?
Now there are 290 licensees. It’s a good question, and one we don’t get very often. There are all sorts of weird applications. There’s a glaucoma monitor chip that’s a cubic millimeter. It’s a pressure sensor, a solar panel, a microprocessor and a radio and a battery, all in that space, so it can be fitted inside the eye so you can be tested for glaucoma. On the other extreme, we’re in a neutrino detector that’s in a kilometers-long chain of sensors, with another sensor every few meters, down in the Antarctic. So we’re in applications that are as small as a cubic millimeter to as large as several square kilometers. Looking forward, one of the ones I’m intrigued about at the moment is with a company that makes concrete. The idea is it concerns networks of sensors that would be embedded directly in the concrete. But you get the feeling that one company is going to pour the concrete and another is going to place the sensors. But this company wants to put the sensors in in the first place. We’ll just pour the concrete with the sensors already there. It’s all about energy harvesting from the vibrations in the concrete. The processors come with little wireless communications [abilities], and use hardly any energy, because the communication is only from one sensor to the next. That one is probably a few years off, but the fact that a concrete company is thinking about this is very interesting.
The next big thing is that ARM chips are coming to traditional PCs running Windows. We’ve been hearing about it for more than a year now, and Microsoft is starting to show Windows 8. Is the opportunity for ARM in PCs real, and is it going to happen?
It’s real and it’s going to happen, and it’s absolutely on track. Obviously, the detailed timeline is a matter for Microsoft and not for us. Metro is happening. It’s a big change to the user interface. They have pioneered Metro in their mobile offering, and you can sort of see where they’re going with it. But Windows 8 is going to be about Metro. That lends itself a little more to tablets in a way that they haven’t been before. That is clearly going to happen. For us and for Microsoft there are two different objectives. For them, it’s about getting a route to support the billions of Internet-connected screens that are going to appear over the next decade or so. Most of them are going to have an ARM processor in them. Without Windows on ARM, Microsoft is excluded from those products, so they need Windows on ARM. For us, a great side effect is getting into the PC world where, outside of Apple, Windows is everything, and it has been inextricably linked to Intel and x86. So now if Windows appears on ARM, we can address those 300 million PCs that are sold each year. And for us, it’s like having an extra 300 million smartphones. It’s certainly nice to have.