Ecosystem vs. IDM: Is AMD onto something?

By Ed Sperling, Editor in Chief -- Electronic News, 3/30/2007

Phil Hester, chief technology officer of AMD, sat down with Electronic News/Electronic Business to talk about his company's ecosystem, why it changed its strategy and why Intel is going to be doubly challenged in the future. What follows are excerpts of that conversation.

Q: Intel and AMD have been slugging it out over megahertz and gigahertz for years. The battleground seems to be shifting though. What's behind that shift?
Hester: It’s really about what the customer wants and what the end customer experience is going to be. About two years ago, which is before I joined AMD, one of the questions the company asked was what it would require to take AMD from a good company to a great company. Part of that transition was about how to deliver the complete solution that the customer wants, whether it’s hardware or software or whatever it is that you’re going to care about in the PC space. As we looked at that, it was clear we had a lot of the assets we needed on the server side. On the client side, it was pretty clear that with the applications gaining with Vista, we needed the same capability on the media side as on the PC side.

Q: What’s the distinction between media and PC?
Hester: If you look at a PC today, a lot of it isn’t how fast it recalculates your spreadsheet or how fast your PowerPoint screen pops up. It’s the visual experience. More and more, we’re running complex video streams. The example of SD (standard definition) video going to HD (high definition) shows up. You’re going to want to integrate the PC into your household at some point and stream content out of it. We thought about what the probability was that we could grow it organically. There are a lot of people who tried to get into the graphics business and stay leading-edge, but there are only two companies that I would argue are on top of the pyramid. That led to us wanting to acquire a company that had that skill. There’s going to be more and more overlap between PCs and consumer electronics in the client space. Moore’s Law is going to continue. As it does, at a given price point you’re going to put more and more transistors on a chip, which will enable more and more capability, and more and more complex software. Along with that the cost of communications will be coming down. So where is the boundary between a high-end cell phone and a low-end notebook computer?

Q: Where is the boundary?
Hester: Other than a keyboard and the display, I don’t notice a lot of difference. Now, with the leading-edge cell phones, you have a [high-definition multimedia interface] so you can plug a 60-inch panel display into your phone. With digital broadcast, it also can become an MP3 player. It’s going to be more limited than your PC in terms of storage, but in terms of capability and bandwidth it’s going to have a lot of the capabilities that today are done by your set-top box or your PC.

Q: Isn’t this like former IBM chairman Lou Gerstner's vision of ‘pervasive computing.’
Hester: It’s pervasive computing and communications.

Q: Intel and AMD have been slugging it out on processors, but in a less-obvious way they’re also slugging it out on business models. Intel keeps its core processes and development internal, while AMD is hooked up with a slew of partners, including IBM. Which is better?
Hester: Regardless of what Intel would like to tell you, people want to be able to pick and choose. Having a monopoly can guarantee compatibility, but it also inhibits innovation and competition. What we’re trying to do is enable an ecosystem and create a platform that allows people to innovate. We continue to broaden that. You need some way for people to build, at a reasonable cost, specialized accelerators. What we’re doing around Torrenza and accelerated computing is a great example of that. We also want a close enough relationship with our partners so that we can go to an OEM and look them in the eye and say, ‘We will guarantee that these things work together.’ That’s what OEMs want. They want a choice, but they also want someone standing behind it so if there’s a problem, who’s going to help them fix the problem and certify that solution?

Q: In the past, OEMs defined the specs and the final product because they were closer to the customers. Is that changing? Are you able to capitalize on the market by being closer to the consumers?
Hester: It’s fair to say that more and more we deal not just with the OEM but with the entire ecosystem up to and including the end users. Now, particularly in the consumer space, what is a consumer going to want on their device two to four years from now? It’s not clear that you can get that information without doing some of that direct end user research yourself. You’ll get it filtered and you’ll get some value from talking to the ecosystem and the OEM, but unless you’re doing prototypes and focus groups you’re not going to be able to get enough information early enough to influence the design cycle. The other thing we’re doing, particularly in the consumer space, is finding a leading partner to go out and do early and exploratory devices in areas that are mutually interesting to us. We create a device in a modular fashion that has exactly what we need to target a market segment, but do it in a way that is flexible enough so that when we learn what that segment wants we can modify a design, sort of like an SoC approach. That is new in how close to the end user we’re reaching. It also gets into the retail channels. What is it that they’re hearing?

Q: So what are you hearing?
Hester: In the PC, it’s validating what we thought was happening—a very media-rich environment. It’s 3-D graphics, video streams. We started thinking about this way before YouTube, when a consumer HD camcorder was $4,000. Now they’re $799, and probably even lower. This whole media-rich, 3-D gaming immersion is turning out to be real. There are more and more capabilities in the device itself, and the device is part of a bigger home environment—whether it’s being able to plug in your 60-inch plasma display to your phone or whether it’s being able to move the content from your PC into this device.

Q: You’re talking about seamless connectivity of every device?
Hester: Yes, and more and more that is happening. From a software standpoint more than hardware, how do you allow people in a friendly environment to move their content around and also comply with all the DRM (digital rights management) issues. The hardware is capable of doing more than it does today, but because of software or licensing restrictions there are some system-level limitations.

Q: How does AMD go beyond its classic market and get into these consumer devices?
Hester: Part of that came as a natural extension from the ATI acquisition. ATI’s consumer side already had a very good reach with the end users and with the customers that are doing exploratory devices. What accelerated with that is the same sort of engagement in the PC space with the [retail] channel and the end customers. It’s a continuum. It’s a continuous growth from people paying attention to this and then also making sure we have a path back to the design teams so we can influence the designs early enough so we can hit the targets.

Q: You’ve mentioned the retail channel. Are you setting up relationships with retailers—particularly in light of the fact that some of them are acting as OEMs for their own brands?
Hester: In general, every major channel that is going to be the end purchase point for a lot of our volume, we’re interested in talking to. That’s not just classical channels. As software becomes a service, if they’ve got a huge data center we’d be talking to the data center. If somebody like a Costco wanted a PC targeted at a specific market, we’d be very interested in having that discussion.

Q: So are you talking to an entirely different group of customers than in the past?
Hester: If you went back five years ago, 99 percent of the requirements came from the OEMs. They’re now one of four or five different sources. Because of the lead time and the complexity, you have to look at the software community. You also have to deal with the Linux community instead of just the Microsoft community. We’re out talking to retail channels and end customers, and more and more we’re having those conversations outside the United States. Five years ago, the majority of discussions were inside the United States. The majority of sales are actually outside the United States, so you have to listen where the sales are going. We have in development centers and customer labs in the emerging centers of India, China and Dubai.

Q: Historically, Intel and AMD have clobbered each other on price, then raised them whenever possible. It appears that AMD has resigned itself to lower prices, given the huge volume of relationships it’s been developing. From our vantage point, AMD’s focus seems to be more on expanding the market than extracting higher prices.
Hester: Historically, Intel tried to do that—bring you in, get you hooked, then raise the price. Our philosophy is to price on value. If we’re doing something and a company doesn’t see value in it, then shame on us. Across the consumer space, what is going to give a better experience? If that requires us to go do an enabling piece of software with the Linux community, the compiler community, ISVs (independent software vendors), and we need to invest resources there in order to create a value to the end customer, we’ll do it. That’s different from when AMD used to do a quick rendition of what Intel did before. Now, more and more, if we add value in the silicon, in order for the customer to see that we’re engaging more of the software stack.

Q: From the outside looking in, the busiest people in AMD these days appear to be in business development. Why?
Hester: This goes back to, ‘good to great.’ If you want to be a great company and a major player in the industry, with all due respect, you’re not going to do it as a monopoly by yourself. You’re going to do it in partnership with others in the industry. It’s not just ego and intelligence. It’s that other people have good ideas.

Q: How does your ecosystem mesh with that of your partner, IBM?
Hester: There are many different answers to that. Where they are integrally related is in the semiconductor process area. We and IBM are absolutely locked together in developing silicon technology, and we have been for about five years.

Q: The original deal was for 65 nanometers, right?
Hester: It keeps getting rewritten. It’s hard to do that stuff right. There are arguably only two or three companies in the world that can continue to do that. But that is a true, interlocked partnership. The fabs are separate. They deploy and use their own fabs, and so do we, but it comes from the common seed that we develop together. We overlap with them on a number of areas. They’re a great customer in the X series. We work together with them with the Linux community. When you look at the Taiwanese ecosystem, whether it’s the chipset guys or the ODMs, we work with them, too. Do we allow them to do differentiated products through Torrenza and HDS? The answer is, yes. Likewise, we do that with Sun and HP, which are seen as their competitors. We’re a supplier of technology to the industry. Our philosophy is that we make the same offer to IBM that we make to Sun or Lenovo. We think that’s the way to play in an ecosystem. The other side of that is sometimes an OEM will say, ‘We need you to go partner with that IHV (independent hardware vendor) to provide a system reference design we’re going to take to market.’ We’re responsive to what they say they want. They may want a Broadcom chip or a Via chip. We listen to that and develop that particular solution for them. Our job is to be flexible, but also to be the single throat to choke if there’s an issue.

Q: That sounds like the old IBM integrated solutions model.
Hester: This is a disaggregated model, though. We’re not trying to be the black hole that everything gets sucked into. We’re an enabler that has a bunch of healthy planets rotating around it.

Q: Still, this loose aggregation of partners and potential partners is one of the most radical shifts in the history of business.
Hester: I’d use the term ‘coopetition,’ like the Unix guys had it in the past. HP, IBM, and Sun worked together to develop open, common standards, but they also competed in the marketplace with products. The same is true with Nvidia. You can argue we are competitors with each other, but if an OEM comes to us and says they want to use an AMD processor with an Nvidia graphics card, why should we tell them, ‘No?’ Likewise, if they come to us and say they want to use an ATI graphics card in an Intel platform, we should encourage that, too.