Q&A: ARM President Talks Windows, Mobile Growth, And Competing With Intel

Thanks to ARM Holdings, the PC chip market may no longer be a two-horse race. In fact, it could get quite crowded in the not-so-distant future.

The British microprocessor company had quite a coming out party at CES 2011. Microsoft CEO Steve Ballmer announced during his keynote that the software giant would release a version of its forthcoming Windows 8 operating system for ARM-based platforms, a first for Microsoft and a move that many see as a major threat to the dominance of the Wintel architecture. And as if that weren't enough, Nvidia announced that it will create a high performance CPU for desktops and servers, codenamed "Project Denver," using ARM's architecture (Nvidia also said it will continue to license ARM's technology to produce future versions of its Tegra mobile platform).

While ARM's profile grew last spring, thanks to Apple using ARM's Cortex A8 CPU design for the iPad's A4 platform, the company has become a dominant force in the mobile device market, particularly in smartphone chips; along with Nvidia and Apple, manufacturers such as Qualcom, Samsung, NEC and Texas Instruments license ARM's chip designs.

Already a force in the mobile device market, ARM is now poised to break into the PC and server market and take on Intel. And with its unique business model -- licensing its chip architectures to various competing manufacturers – ARM, which was not too long ago a little known play in the chip space, could shake up the microprocessor industry. CRN spoke to Tudor Brown, president and co-founder of ARM, during CES 2011 and got his thoughts on Microsoft's blockbuster announcement, his company's enormous growth in the mobile market, and going head to head with Intel.

id
unit-1659132512259
type
Sponsored post

What does Microsoft's decision to make an ARM-compatible version of Windows 8 mean for your company?

It's an interesting question. A lot of people are asking the question, "Does this help ARM or does this help Microsoft?" And of course, the reality is it helps both. But if you go back five years or whatever when we were having these kinds of discussions with Microsoft, their response was, "Well, fine, but what's in it for us? How does it help us grow our market?" Because at the time, the PC market seemed just fine and Microsoft didn't need any help at all.

What they're now saying is a very different message: Microsoft needs to move into the future, and the way to do that is through ARM. So the way I see this unfolding is, computing is changing and it's splitting into cloud services. The PC industry has evolved; it's not a very exciting industry, as far as I'm concerned, and it's pretty static and it's not really growing. But what's happening is, the growth is coming from all of the connected devices, and that's supported by the cloud. So these two things, PCs and mobile devices, do go hand in hand. And of course, Intel has total dominance in PCs and has total dominance the servers powering the cloud.

And ARM has…well, not total dominance, but a pretty strong presence here in the mobile device space. And now Intel obviously saying it wants to come into that mobile device market, as we're saying, "No, I don't think so." And we've opened up ARM coming into the server and PC space now, which we don't think is particularly exciting, but there's no harm in it. Plus, it will help bring down the price. Whereas netbooks were a lower cost alternative to notebooks -- and not particularly good alternatives -- you'll see a lower cost notebooks based on ARM. It's inevitable because ARM and ARM partners can produce cheaper systems than Intel. That's a fact. So you will see lower cost PCs soon.

Next: ARM's Low Power Proposition For Servers, PCs

There are a lot of devices now in that mobile segment. Does that give ARM an advantage? And where does the server market come into play?

Well, looking at the growth of computing, we've been sort of stuck here in the desktop Internet era, and now we're moving into the mobile Internet era. And it's a lot bigger market, and it's lot more interesting. The future is about the Internet. And these two forces, the desktop PC and the mobile device, have been going on for some time. And I see this sort of bifurcation of the computing industry. The computing industry isn't going away. The PC market is obviously going to continue. The market isn't dead, and Intel certainly has a future. But to me, the excitement is in the mobile devices market. And that's matched by growth in the server market. And whereas as ARM isn't targeting the server market as a major growth area, we do think that something has to change there because these two markets, mobile devices and servers, are connected through the cloud. And we do think ARM can offer something there in the future.

So how will ARM's offering be different than the competition's existing technology? How do you plan on addressing that server market in a way that's different than Intel?

Let's look at what's happening in the mobile market; as these devices become more prolific, you get more demand on Intel-based servers. Whether it's GPS or social networking applications, all of these devices in some sense are only interesting if they're connected [to the Internet]. So there's absolutely no question that there's growth in servers -- it's guaranteed for the next 10 or 20 years. But the problem is, these things are terribly inefficient from a power consumption point of view. And the analysis is that if we do nothing about this and just these servers grow mindlessly, then by 2020 there will be more energy used in these servers than in aircraft travel. Already today, servers account for between 3 and 4 percent of all U.S. energy use. It's an astonishing amount of energy that's being used in data centers, and it's only going to grow. So the only solution is to make lower power consumption servers that are more energy efficient.

Now, to be fair, Intel has woken up to this fight and is doing a little bit to improve the efficiencies of its servers. But we don't think they're doing enough. And the total cost of ownership for servers isn't just the purchase price; it's also the cost of running those server farms. And that affects the density of those farms; it affects how many servers you can actually have and how powerful those servers can be. So we see that in the future, servers have to go lower power. Naturally, we think ARM's technology is going to be suitable for that in the future. And there needs to be some work done to get there.

But we're absolutely not pumping up our road map for servers necessarily. But what we're saying is, look, the growth in connected mobile devices absolutely guarantees there's going to be growth in servers as well. And we think something can be done about that server market to make it more efficient. Some of our partners are experimenting with putting ARM technology into servers because they can see the problem and they know that something has to change. And I'm sure just as it did for PCs, Intel will bring out lower power products for this space. But it probably won't happen fast enough.

Next: ARM's Speed Advantage And Partner Community

So ARM can move much quicker than Intel?

Well, our partners that can move quickly. They can move quickly because they've got nothing to lose -- whereas Intel has a lot to lose. That's part of it. The other part of it is, there's a natural competition within the ARM partner community. Intel has sort of a monopoly, but ARM enables a natural competition between its partners, and that means that they're pretty aggressive. Another thing that's happening is, these server farms today tend to be very big investments, very formal deployments, but we'll probably see a move to more ad-hoc server deployments and more distributed environments. A lot of these servers are using a tremendous amount of power, but they're not really doing anything. They're not really being utilized to their full capacity. I liken it to having a bunch of V8 engines sitting in a traffic jam. So a more ad-hoc, flexible system will give you dramatically lower energy costs. So that's what we're looking at. It's not really for us to develop server chips -- it's for our partners to develop server chips.

So how quickly will this happen? When will we start to see ARM architectures in servers?

Well, of course, it's not just the hardware; there's a whole lot of software and infrastructure that needs to be in place, too. And there's a sort of a catch-22 involved here because you need certain hardware to get the right software ported and so on. So that's going to happen over the next couple of years. The conversations are happening now. The investigations and experimentations are happening now, and people are waking up to the problem now. But the solutions are a couple of years away, if not more.

Besides energy efficiency and power consumption of the chips, are there other areas in the server market you feel aren't being addressed?

There's a whole lot of things that come together in this market. It's not just the microprocessor's power. It's also the memory technology, which is obviously improving all the time in terms of capacity per watt. There are also hard disk drives, which probably over time will move more toward solid state storage. And then there's cooling, too. So there are a lot of things that can be done if you put power consumption of the server as the design criteria instead of just performance. And this is an area that ARM has been familiar with for a long time. It's not just about getting maximum performance; it's about getting the power consumption down for average performance.

Next: Nvidia's Tegra Platform And The Tablet Boom

Do you have a specific approach in mind for attracting new partners in the server market, and is there a specific type of partner you want to bring in?

Well, here's an example: it's public knowledge that we've made an investment in a startup called Calxeda (formerly Smooth-Stone) in Austin, Texas, to deliberately prime the pump, if you like, to get that market started and do something that is focused on the server market. And it's not on ARM's P&L and isn't part of ARM's core business and it's not really what we should be doing. But we're supporting them and encouraging them, and we'll continue to do so because we want to get to that end result.

And again, there are other more traditional ARM partners that do a lot of stuff in the mobile space that are also in that backbone infrastructure space. Look at the announcement from Jen-Hsun [Huang] and Nvidia -- he's talking about high-performance computing. What exactly does that mean? You can read lots of things into that. Clearly, some of these servers do need high performance, so there are a lot of possibilities.

Speaking of Nvidia, can you talk about how that partnership has grown and what it means to ARM in light of its recent announcement?

Nvidia has been an ARM partner and licensee for a long time with Tegra and its mobile products. And Tegra is being adopted by quite a lot of tablet makers, and that's great. We're very happy Nvidia has been successful there and we wish them continued success. Now the announcement this week was about Nvidia focusing on a new and different market area using ARM technology. And in that sense, assuming they are successful and are able to carve out a certain market share, then that will help us grow the total ARM pie significantly. The fact that Nvidia is making a huge engineering investment and being very serious about this new market has got to be good for us. Nvidia isn't stupid; those guys are very bright, and they're very competent when it comes to turning out complex chips on a very tight time frame. So if they're doing that around ARM in a new area, that's a huge plus for that new ecosystem we discussed before.

Let's talk about tablets, which are everywhere at show. How do you see tablets affecting the mobile market?

Well, they're ultra-thin, right? That's an attribute that fundamentally comes from lower power consumption. It's very simple and easy to see why the two things are related; you spend less energy, you have less thermal engineering. A lot of the engineering that goes into laptops and notebooks right now is thermal engineering; it's about how you get the heat out of these things. So if you have lower power consumption to start with, then you can have much simpler designs and smaller form factors. So thin has become the results from the drive to lower power consumption. It's similar with TVs, too. You have ultra-thin TVs now that have to have lower power and more energy efficient platforms.

I've been very interested walking around this show this year and seeing all the green technology here. Many companies are really pushing their green credentials this year, which is great. ARM is very much aligning itself with that effort. We want to help drive it. And the drive to lower power isn't just about form factors and making devices thinner. It's also about usability -- making these mobile devices run longer and extending their battery life so you don't have to keep charging them every couple of hours.

Next: ARM's Competition With Intel

Going back to Microsoft and its announcement, there are quite a few people in the media that believe ARM has effectively ended the Wintel era and changed the game completely. Is the media making too much of this?

I think it's significant, without a doubt. I think the temptation of the press, and maybe people in general, is to look at the market and say that ARM is now going to be in the desktops and servers space. And there will be some of that. But that's not going to get me out of bed in the morning. Because as I said before, this PC market is a stagnant market and not very exciting. What's relevant is that the PC market enables more of these mobile devices, and that potentially creates an even bigger core market for us. I have no interest in taking market share from Intel in their core market. They can have that market. What I am interested in doing is stopping Intel from taking market share in the mobile space. So going back to the Windows announcement, Microsoft and ARM are helping each other. The reality is, Microsoft sees an opportunity for growth market share in that mobile space with ARM technology.

Is there a concern, looking at the competitive landscape, that Intel will gain on you in your core mobile market?

Intel will certainly gain some market share in that mobile space. They'll have tablets and they'll sell those tablets, no question. But I still think that Intel's technology is not really appropriate at this point for [the lower power consumption mobile device]. However, let's assume for a minute that Intel suddenly develops products identical to ARM's. I still think we are in a very strong position because in this mobile market, you're talking about billions of units and logicially, you have to have multiple companies in that supply chain dedicated to this growing market. And the ARM business model allows that; it enables choice of the OEM, and it enables competition, which creates lower pricing. So those are the real attributes that have enabled the huge variety of products in the mobile device space. Look at smartphones; I've said the PC market unexciting and they all look the same -- the only difference is the color of the case.

But you look at smartphones, and you have different form factors, user interfaces, and features. It's a much richer market, and that's fundamentally created by the ARM business model, not ARM technology. So maybe the only result of all this ARM server talk is that Intel is forced to bring out better low-power products. And that would be good result for the world, right? And I'd thank myself for that! [laughs] But I think ARM will go a lot further than that. It's a ways off, but I think we'll make some real inroads.