The folks at ARM, the Cambridge, England-based designer of low-power processors, are kind of greedy. They pretty much own the mobile processor market, through their dozens of licensees who make chips for everything from the Apple iPhone to the latest LG curved smartphones.
But they also want to own the Internet of Things, or smart and connected everyday devices. Jem Davies, an ARM fellow and vice president of technology, told me about the company’s plans to get its chips into Internet of Things gadgets. In fact, ARM chips are already in a ton of smart watches, coffee makers, helmets, and all sorts of other gadgets. Intel, the world’s biggest chip maker, is definitely investing heavily to take this market for itself. But ARM isn’t about to let that happen, according to Davies.
We caught up with Davies at the recent 2015 International CES. Here’s an edited transcript of our interview.
VentureBeat: I’ve seen a lot of clones of the Apple Watch already. I don’t know if you’ve seen your stuff in any of these Chinese models coming out already?
Davies: We have. Most of the smart watches and things like that have been based on ARM CPUs and GPUs as well. The first ones have come out using existing chips, mostly. We only get to know when a chip starts. We don’t necessarily know all the applications a chip goes into. Sometimes you know a great deal more about what my graphics chip goes into than I do.
VentureBeat: What sort of chip is going into these watches?
Davies: Watches are usually the Mali-400 MP1. We worked closely with Google on the specification for Android Wear. That mandates OpenGL ES2. It very specifically doesn’t need OpenGL ES3. Mali-400 is the perfect fit. It’s much smaller and lower-power than rival chips. It’s cleaning up in that space.
VentureBeat: Does it seem like a smartphone around 2007, that kind of capability?
Davies: That’s about the size of it, yeah, if you mean a smart watch as opposed to a fitness center. The fitness centers are very Cortex M, no graphics, that sort of thing. But a smart watch is a first-generation smartphone type of design. If the display is small enough, you can get quite a cut-down GPU.
VentureBeat: If you look at things that are shipping well in this Internet of Things category, or whatever categories you measure, what do you notice is actually selling?
Davies: In terms of volume there is no market in the world like mobile. Mobile, which includes tablets, is shipping hundreds of millions of units for us. It’s probably outshipping everything else 10 to 1. They’re shipping hundreds of millions of units of tablets in China. The U.S. analysts are not necessarily counting those very accurately. I’ve had to correct some U.S. analyst numbers, because they don’t see what’s going on in China.
It’s crazy. They’re doing 7-inch tablets for $35. Now, they’re not that great. They’re not a high-quality device. But they work. I bought one as a hi-fi controller, for my Sonos hi-fi. Why not? It’s probably not going to last very long, but okay, I’ll buy another one.
Tablets have almost become disposable items now. Our CEO and I did an investor breakfast yesterday morning. He was talking about seeing customer presentations where people hand out tablets to the audience. Do they get all of them back at the end? Probably not. But what you lose in the cost of it, well, whatever. It’s opening up new uses that are quite interesting. I obviously do a lot of flying. At the airport, you used to have a man holding up a clipboard with your name on it. Now it’s a tablet. Didn’t see that one coming, but it’s interesting.
They’re also becoming quite commonly used cameras. People like the big viewfinder. It doesn’t work for me – I want to hold something up to my eye and look through a viewfinder – but maybe that’s just my age. An awful lot of people now are holding up tablets to take pictures that way. That requires a lot of image stabilization, because hands shake.
The next biggest market for us is digital TV and set-top boxes. We’re in 70 percent of the market in GPU-enabled DTV, which is pretty much all TVs now. We used to have to use the “GPU-enabled DTV” line, but soon we won’t have to say that. We’ll just say DTV. The Samsung, LG, MediaTek, NovaTek, all those guys are using Mali. We have a clean sweep across there. Set-top boxes, less penetration, because you have Broadcom. They have their own GPU. But in DTV we have a very good swathe.
VentureBeat: I’m trying to come to terms with some of the Internet of Things stuff at the show. You take a $20 dog feeder and you make it a smart connected device. It’ll tell you all these analytics about your pet. Then it becomes a $250 dog feeder. I like the idea of a smart dog feeder, but I don’t like the idea of a $250 dog feeder.
Davies: You don’t, but the manufacturer does. [laughs] We mustn’t get too fancy about ourselves. ARM is a components buyer. We supply technologies to people who have clever ideas. We’re feeding that.
VentureBeat: Do you think the technology is there to fairly rapidly make that something like a $30 dog feeder?
Davies: Yeah. And there, time to market is king. The guy who gets the product for that golden window of three to four months charges a lot more than anybody else. One thing we’ve put a lot of effort into is the speed with which people can integrate our IP. We deliver them a bunch of software, and they handle and produce a chip. If that process can take four weeks rather than four months, those guys have just made a lot of money.
It’s the hidden quality angle of our IP. Everyone measures the power use or the silicon area or the bandwidth. That’s all good. We concentrate on those as well. But the secret weapon, what we sell to people, is that it’s easy to integrate. An awful lot of software is already pre-integrated. We have a complete Android stack. Our graphics, video, and display processors are all pre-integrated into the Android stack. That’s a chunk of work. We’re putting a lot of work into getting people started as quick as possible.
VentureBeat: Some of these devices are pricing for a luxury market that I’m not sure exists. I saw that skateboard-like, Segway-like thing. It was $1,800. Would you buy that?
Davies: It opens up both markets. It opens up a commodity market. It’s now cheap enough to put a processor in everything. That’s not going to significantly affect the price, so it opens up a whole commodity market. But also, people are going to say, “Well, I’m going to create a luxury segment. I’ll take this capability and market it as a hugely valuable aspirational something-or-other.” That’s interesting.
VentureBeat: Give it the Gucci treatment or something?
Davies: You can pay 10,000 pounds for a smartphone. You get a diamond-encrusted smartphone. You have to applaud that. Am I going to buy one? Hell no. But it’s that sort of creativity and diversity of what the ARM partnership does that I think separates us from the likes of certain companies down in Santa Clara.
Our partners are doing some wacky stuff. Some of it works and some of it doesn’t. You see certain product categories capture the public imagination and they sell tens of millions or hundreds of millions. You and I might have thought it would never work, but somebody does it and it’s a fantastic success. Do you have a phablet?
VentureBeat: Yeah, I do.
Davies: I looked at phablets and said, “What is the point in that? It’s too big for a phone and too small for a tablet. That’s stupid.” Lo and behold, people went out and made them and have sold 100 or 200 million units last year. If I’d been predicting that, I’d never have done it. It taught me a valuable lesson, which is that what I think doesn’t always count. We should stick to what we’re good at, which is the component technologies that enable other people to have clever ideas.
VentureBeat: There’s that fine line between clever and crazy. Price has a lot to do with where that should be.
Davies: You have the reverse psychology thing. If you sell it for $18, people aren’t interested. If you sell it for $1,800, suddenly it’s incredibly attractive.
VentureBeat: What do you think are the emerging markets that you’re winning? In some ways Intel has finally stepped up. Everybody in the world has stepped into the Internet of Things. Is ARM still coming out on top?
Davies: So far the Internet of Things is absolutely running on ARM. Pretty much every single device you come across is based on ARM. No one else has made significant inroads yet. Can we keep up that market share? Well, it’s an open market, an open playing field. We have no right to a huge share. If they produce products that are useful and attractive, they’ll do well. But at the moment, wearables and the Internet of Things pretty much run entirely on ARM.
VentureBeat: What about your push into servers?
Davies: That’s a slow burn. We said last year that by the end of 2014 you would see ARM-based servers being shipped in commercial systems, and you do. They’re popular. HP is currently out of stock. You’ll have to go on a wait list. That must be selling well. I haven’t seen the numbers yet, but it’s doing well.
We see that as something that’s going to ramp up significantly. But it’s not like mobile and Fitbits. It’s not going to ramp up in three months. It’s going to ramp over some years. You’re talking about very conservative institutions buying these huge data centers. They have replacement policies. They’re retiring their kit over three- and four-year cycles.
The critical part there is not just the ARM low power, but the fact that we’re enabling diversification. Rather than saying, “The answer is this server or that server,” we say, “Here are our processors. Here’s our technology. Go build what you think you can make a successful product out of.” People are building networking servers. People are building transaction servers, web servers, base stations with server capability, a whole variety of product categories that previously weren’t possible. I think you’ll see that as one of the great advantages.
VentureBeat: What about the stories like Xiaomi in China — products that are designed just for China that also become giant. Are there a lot of examples of things like that?
Davies: We’re very well-placed in China. One of my colleagues on the senior management team of the graphics division is based in Shanghai. He’s been there for two and a half years. We have a significant engineering presence in Shanghai as well. We’ve been in the right place at the right time as companies like Xiaomi — we’ve got in at the start with these companies. They’ve been licensing quite a lot of technology from us because they have no particular legacy.
A greenfield account just says, “OK, I could build that, or I could buy that.” Oftentimes they say, “Yeah, I could build that. It’s a bit more expensive, because it’ll take me longer than just buying it,” but the time-to-market advantage is huge if you choose to license from us. Knowing that they can license it from us, that it’ll work, that the integration and the software will be easy — that’s something that has tended to be successful for us. If you look at the 70 licensees we have, a huge number of them are in Asia.
Xiaomi is an amazing company. The speed with which they work is terrifying. They’ll ring you on a Sunday night and they expect an answer before Monday.
VentureBeat: The infrastructure for the Internet of Things, is it already solid?
Davies: No, I think there’s a lot of change to happen there. Low-power Internet, low-power wireless, low-power communication protocols. There’s still a lot of work to be done. That’s one of the reasons why we bought a company called Sensinode, to help establish low-power Internet communications protocols. We saw that if you really do want something that’s going to be powered off a button cell, nailed to the wall, and left there for five years, you can’t do 802.11ac. It’s just not going to work.
VentureBeat: I saw some companies starting to do batteryless devices, trying to generate their own energy.
Davies: When you get the processes down to half a volt, things like that, that starts to become possible. It’s still hard.
VentureBeat: A lot of these things aren’t moving, too. You can’t harness motion energy.
Davies: The problem is, we all know you can have a self-winding clockwork wristwatch. There is energy to be harvested. But with that sort of smart watch there, it’s using quite a lot of power still. But it’s getting there.