CES 2016 drew more than 170,000 attendees.

Above: CES 2016 drew more than 170,000 attendees.

Image Credit: Dean Takahashi

Question: I wonder if we can connect this to some of the larger tech trends at CES. We have half the market declining in smartphones and PCs, the established categories. But then these new ones are bubbling up. They almost seem like custom businesses from your perspective, like robotics or drones or wearables or VR. I wonder if you then have to think of semi-custom businesses in each of these categories.

Su: They’re very synergistic. It’s the same types of components that you need. You need computing horsepower at the right performance per watt. You need visualization capability at the right pixel density. Whether you’re talking about drones or PCs, I’m not sure the CPU is all that different. There are a lot of things that are similar in that view.

Maybe the difference in how AMD is approaching it is — look, I like the traditional business. As much as we talk about it as a declining business, it’s a technology driver. It’s the volume driver that affords the ability to develop all that technology. And then yes, we will veer off into a VR ecosystem. We may veer off into a robotics ecosystem or an AR ecosystem. But the basic technologies are very similar. That’s the way we’re approaching it.

Question: You mentioned the traditional business. It seems like there was a time where AMD was talking a lot about increasing the portion of revenue coming from embedded and semi-custom. There were some goals on that and a big push. It sounds more like you’re emphasizing desktop APUs and graphics now. Has there been a change in strategy there?

Su: It’s certainly a change in what we talk about to you guys. When we were 95-5, we needed to make sure that people understood that there’s an AMD outside of the PC business. We’ll end 2015 roughly 50-50. That’s a good place to be.

I want to emphasize on that point that the traditional business is still a great opportunity for us. When you look at our share in certain segments of the market, we’re underrepresented for our capability. I view that as an opportunity for growth in 2016. When you look at the custom business, I love that business. But the custom business could not be there without a strong IP portfolio from the traditional business.

Question: Some of what you described also sounded like where you’re different from Intel and Nvidia. Nvidia is putting a very big focus on cars. Intel has the Internet of Things. They’re customizing their line for all these brand new businesses.

Su: It is different. Our focus is on the basic technology and the IP. That’s a place where we have a lot of room to grow. You’re right. We’re not talking about self-driving cars.

Question: Do you think that’s a benefit?

Su: I think it is. We’re very clear on: When you ask what we want people to think about when they think about AMD, I want you to think about AMD as a technology powerhouse. It’s delivering great products in the segments we’re in today. There’s more than enough opportunity, given our size.

Question: Nvidia still likes to put out a big part for graphics and gamers, staying on that bleeding edge, but you guys have all the console business. Is that an insight you had a while back, that the console business would not be on that bleeding edge? Or do you have to have that high end to get other business?

Su: I don’t think you need the high end to get in the console business. But I think you need leadership performance per watt capabilities. Any APU, any discrete GPU, will benefit from performance per watt. We were behind. We’ve significantly accelerated where we are with our new technology.

From the standpoint of the king of the hill product, that’s always important in the channel business and the add-in board business. That was where we targeted Fury with our initial high bandwidth memory. We’ll continue to have a king of the hill part that competes in that area.

Question: Is there any time frame for that? Is there a 2016 king of the hill product?

Su: Give me a few more weeks into 2016 before I have to answer that question. We will update the entire portfolio in FinFET, including the king of the hill. But timing will be updated.

AMD Opteron 1100 is a long-awaited ARM-based processor for the data center.

Above: AMD Opteron 1100 is a long-awaited ARM-based processor for the data center.

Image Credit: AMD

Question: I assume some of that is also dependent on the development of second-generation high-bandwidth memory (HBM)? How is that coming along?

Su: It’s coming along well. We learned a lot in the first generation of HBM as far as the manufacturing and the bring-ups. HBM 2, we’re very much committed to it, committed to this 2.5D architecture. It makes a lot of sense going forward.

Some people were expecting that we might update the entire road map immediately with high bandwidth memory. That was never the plan. There was always an expectation that there would be high bandwidth memory first at the high end to get the technology out there, and then as the cost points get better — it’s still very costly.

Question: Is graphics going its own way more?

Su: I wouldn’t say it’s going its own way within AMD. It’s not second fiddle. If you think about the evolution over the last five or six years, with the integration of ATI there was very much a focus on APUs as the center of the universe. APUs are very important. Don’t get me wrong. If you look at the bulk of our revenue, it comes from APUs. But the graphics technology is so fast-moving and has so much leverage across the APU business, the discrete business, as well as just pure intellectual property, what I wanted was more of a focus. You have two first sons, is the way I would describe it.

Question: But not necessarily positioning it to be on its own?

Su: No. I would say we’re positioning graphics to be in as many applications as possible. Some of them will be paired with AMD CPUs. Some of them won’t. That’s OK. That’s what we’re doing.

Question: Product and technology have been one thing. The financials have been something else. At one point in time you were an integrated manufacturing company and that’s gone. You’ve had step after step after step. The latest one last quarter was the joint venture with Integrated Test. How many more levers do you have to pull, or do you have in your pocket to pull? It seems like year after year goes by and little pieces of AMD just keep getting partitioned off or sold off or partnered off so you can raise capital to keep going.

Su: There’s no question that the financials are the judge of how well we’re doing, how well the technology is doing. We’re very clear about where we add value to the industry. That’s in the technology. Relative to the China joint venture we did in terms of the back-end assets, I think it was a great deal. It allows us to leverage the asset from an overall standpoint. We weren’t completely utilizing the asset. Being able to partner with someone who can utilize it brings value into the equation.

I don’t view that as a pure “let’s sell off something to raise capital.” I view that as becoming a more focused company. In that sense, being able to use a number of different manufacturers that are expert in the area is better than trying to do everything yourself. That’s what we’ve done a bit more as AMD over the past few years. We’re identifying what makes us special and where we should leverage other external sources.

The overall expectation is that 2016 will be a better year from a financial standpoint than 2015. We have chosen to bet on technology. As those products come out, the expectation is that the market share, the revenue, the overall operating profits will follow. But that’s yet to be proven. That’s our job, over the next 18 months, to do that.

AMD Polaris

Above: AMD Polaris

Image Credit: AMD

Question: If that scenario doesn’t come to fruition, what’s the fallback strategy?

Su: Well, I intend to make that scenario come to fruition. But I think you should also think about AMD much more broadly. The technology assets we have are unique in the industry. One thing we’ve also started talking about is our ability to monetize IP outside of traditional product sales. I believe we should do that in any case, because there’s value to it.

My philosophy is, I can’t make every product that can possibly use a high-performance CPU and graphics. Why shouldn’t I enable others, in a positive fashion, to leverage AMD IP in more places?

Question: It seems like Nvidia is trying to do the same thing.

Su: The approach of direct patent licensing or patent enforcement is one avenue. I think we have a much more partnership approach in terms of how we deal with things. It’s around technology enabling and co-development. Almost an extension of our custom model into more of an IP-enabling model.

Question: What does that mean in the real world? By talking about co-development, do you mean partnering, or just taking another chip company and licensing them your graphics technology to do what they will?

Su: That could be one scenario. It could be licensing graphics IP. It could be joint development on CPUs. It could be any of those avenues.