But these are tough times for the one-time Silicon Valley highflier. The company still struggles with large financial losses and has announced layoffs. All the uncertainty has led some analysts to speculate that Sun is destined to become an acquisition target.
Fowler is having none of it. He notes that Sun's server revenue was--a period during which most server makers suffered declines in server sales. Also, the company's sales of Advanced Micro Devices-based servers grew. And so what if Intel has a new server chip coming out? If demand exists, Fowler suggested, Sun might even adopt the chip. He recently spoke with CNET News.com about these and other issues affecting Sun and the chip business.
Q: What's the benefit of combining the x86 server group with the Sparc group?
Fowler: What's been happening is the two groups have been doing things like sharing mechanical design and doing some aspects of the systems work together. It will be a more clear dynamic when it's all one group. For example, service management software can all be shared very quickly and easily between the systems. The other thing that we can do is, from a portfolio management standpoint, is continually adjusting the family of products to maximize our growth. As you probably know, we grew 7.6 percent last quarter.
There are a lot of opportunities in other areas like the factory operations and supply chain. The x64 group had a separate operations team.
I'd like to be your worst-case scenario customer for a moment. So I'll say, "Well, you lost
Fowler: I think the important thing about looking at sort of paper losses and GAAP losses and all that is that we continue to maintain a strong R&D portfolio and a strong cash position, which is sort of what drove the 7.6 percent growth in the last quarter. We don't have any debts; we have an extremely strong cash position; and we've got a very stable and increasing revenue base, so we're in actually a pretty good spot from that regard.
Right, but you are laying people off. You'll obviously have fewer engineers. Do you worry that you won't be able to continue to innovate at the same pace as IBM or even Hewlett-Packard?
Fowler: I think we've obviously been through a very long history competing with IBM, which has a much bigger work force. If you'll look at their portfolio, there are a lot of businesses that we are in fact not in, so it's not fair to jump or take a direct comparison.
But againstand the X series products and other parts of the stack, obviously we have Solaris and a lot of opportunities to compete against IBM, despite the fact we're smaller in scale. Remember that IBM has a huge number of people which are actually part of the service group.
In the case of HP, of course, it's a very different profile of a company. Competing against HP obviously involves emphasizing the areas of portfolio where we are different: the Sparc portfolio,.
Aggravating customer question No. 2. So you are selling a lot of x86 servers. But why should I buy them from you guys? There are a lot of people out there with them.
Fowler: When we develop our products and, of course, we're on the cusp of announcing a whole collection of more products, we work basically on five principles. The five principles are: performance; efficiency--and efficiency relates to both our size and power performance; No. 3 is reliability, which is of course the whole availability track; and then manageability, which is how well can you integrate it into an environment of hundreds and thousands; and then the fifth is what I call longevity.
These are the big five: performance, efficiency, reliability, availability, manageability and longevity. If you look at our x64 products, we have the first four very much nailed. We are fast and extremely power-efficient. We're manageable and have a lot of reliability features, so that makes us clearly a top tier-one vendor.
If you look at our x64 growth rate, we are seven quarters in with really high double-digit or triple-digit kind of growth rates. I can't talk about the growth in the current quarter, but the competitive landscape hasn't really changed. You don't get eight quarters of growth in that rate by accident. People obviously like what we are doing.
The last point I want to talk about is longevity. You're familiar with longevity in the Sparc line. You can install an UltraSparc IV machine and keep upgrading it over time. The products that we are going to introduce next will really bring that to x64. You'll be able to incorporate our blade server into your infrastructure so that future processors, future memory, future I/O (input/output) and other aspects of the system can be continually changed over time . This is quite new for the x64 area, and we think it will really resonate with customers.
The other thing, obviously, in terms of growth is Solaris and x64. We have a majority of Solaris licenses which are actually on hardware other than Sun.
Is that Solaris x86 or that's Solaris everything?
Fowler: Solaris x86. If you look at it numerically, the majority of that is actually on HP hardware. So when you ask why would people want to buy things from me, well people have become interested in Solaris, and they may want to also purchase hardware over time from Sun because obviously we're optimizing and supporting them together.
Fowler: It's free. You download it from Sun. You have access to base level updates for security patches as well as other information. If you want to have advanced support, then there is a service plan for it.
What percentage of your revenue is x86?
Fowler: It's still small. If you look at comments (former CEO) Scott McNealy made, he said that the server part of x64 is a nice little $400 million business. I think it's more fairly classified at around 10 percent of our top line, but growing rapidly.
Everyone is queuing up for a
Fowler: I think Woodcrest is obviously a much more competitive product from what we've seen so far, but it remains to be seen how well it does in a variety of customer applications. Just like anyone else, I am always very cautious about what I call bench marketing. Not that we would do that of course.
But it (Woodcrest) certainly increases the competitive profile. We've always been in a situation where we want a chip we believe to be the best product in the marketplace, and we don't have any issues with Intel Woodcrest and other forthcoming processors becoming things that are highly competitive that customers would want. We wouldn't have any issues adding them into the product line. For us, it's a neutral thing.
So you guys would be open to the idea of actually carrying a Woodcrest server?
Fowler: Of course. We don't have any kind of strange agreement (with AMD) so AMD knows that...I think the more interesting thing about Woodcrest is that it puts enormous amounts of pressure on the whole Itanium thing. As Intel increases the capability of their x64 Xeon-type platforms, it sort of makes that whole Itanium thing look more and more...interesting.
About a year ago, you and I spoke, and I asked you how many people use your x86 servers as 64-bit servers. At the time you said about 70 percent are actually using them for 32-bit apps and 30 percent for 64-bit ones. Has that changed at all?
Fowler: I think it's probably gone up a little bit. Obviously people are using Solaris, and Solaris is only 64 bit. On Linux it seems to really vary. In Windows 64 I don't have a real sense of it. So I think we're up a bit from that, but not necessarily some huge amount.
Now we can switch over to the Solaris side of things. How are sales going there?
Fowler: So obviously we can't talk about the current quarter that much, but the 7.6 percent growth that we had in the last quarter was largely driven by Sparc, because on a dollar it's much higher. The really strong part of the product line that drove sales is UltraSparc IV. But the other thing that's growing rapidly is the Niagara products.
Looking at the overall server market, revenue is still relatively flat. Is there anything out there that will move the needle?
Fowler: There are a couple of macroeconomic trends that potentially change the server marketplace. When I look at telecommunications going to IP and then consumers demanding videos delivered at high bandwidth to everyone's home, it can certainly drive significant change in the computer market. People are talking about Sarbanes-Oxley. I think there is an opportunity here for some big changes in the storage marketplace.
Everyone is talking about the third world, so that's obviously a big thing. Once you get people on the network, we're going to drive exponentially an overall demand. I think that in general the cost of individual aspects of computing will continue to come down. Servers will get smaller and cheaper.
Does energy consumption come up a lot in conversations with customers?
Fowler: Every single meeting. Everyone has figured out it's a big issue. What I found is that a surprising number of customers get the three basic things mixed up.
You have power consumption, which is how much we use from the socket. Then you have efficiency, which is how much work do you get done for what. And then there is power density, which is how much heat do you dissipate in a limited space. Those are the big three issues, and the customers are just sort of now getting to understand the difference between the three and how it affects them.
What's the metric people should be looking at?
Fowler: I think there are two that people should look at. The first is power efficiency, because that relates to OpEx (operating expenses), and then the second one that people ought to look at is power density. Power density is the amount of heat you dissipate in a small area, and the reason you have to look at that is because that relates to your CapEx (capital expenses). How good is the cooling in your data center? You don't want to rebuild the data centers frequently, so if you built the one level of power density you want it to last for four or five years.
When I talk about longevity, the blade servers, what we're going to do is come to people and say "If you built your data center for power density, you will be done for years, because we're going come in and sit within a power envelope. That's the big part of our longevity discussion, which is how do you minimize your repeated CapEx expenditures.