Staff Writer, CNET News.com
SANTA CLARA, Calif.—Sun Microsystems is a company that's made good use of its visionary impulses to survive against bigger rivals. So it's fitting that Jonathan Schwartz is the company's No. 2 executive.
Schwartz has a reputation as an ideas man, but now he's got to turn those ideas into reality. For Sun, which is still working to reverse a three-year revenue slide, that suggests a period of intense change.
Schwartz has climbed Sun's ranks since the company acquired his start-up, Lighthouse Design, in 1996. On the same day in April 2004 that Sun announced a detente with Microsoft, along with its third major round of layoffs in three years, he was promoted to president and chief operating officer.
Since then, Schwartz has pinned much of Sun's turnaround plan on software. So it is that the company's salespeople are no longer compensated simply for selling hardware. Meanwhile, the Solaris operating system is becoming open-source software, and adding a Sun database has become a real possibility.
But Schwartz intends to move beyond software into the Sun Grid—blocks of computing power that he believes will reel in new customers who previously bought and operated their own equipment. Schwartz talked with CNET News.com at the company's annual analyst event last week.
Q: Over the last year, the economy has been recovering and the server market has recovered from its revenue declines. But Sun hasn't seen nearly as much of that growth as IBM and Dell. If your sales pitch is so compelling, why is it that the revenue numbers haven't been going up to show it?
A: I think the average revenue per unit has been going down, but in terms of unit volume, I think our share has actually been growing, quite to the contrary of what you were talking about. If you look at the growth, for example, on the x86 (processor) side, we are now the market leader in Opteron servers over HP and IBM. Who would have predicted that? Where we haven't been seeing the growth necessarily has been in the high-scale data center systems, but that's been reflective of the industry as a whole.
I recognize that unit shipments are significant and understand your argument that volume begets volume, but revenue also is relevant. Do you believe that at some point the revenue will pick up, and if so, when?
Thank you for the offer, but I'm not going to make a forward-looking statement. The unit volume numbers are important to us, the revenue growth is important to us and we've grown half-over-half from last year to this year. Certainly we're prioritizing growth as a company, and the leading indicator for us of growth is the adoption of our software platforms. So we're seeing more Java-enabled handsets go out into the world, seeing more XM Satellite Radio clients go out into the world, more Solaris downloads off of the Web site—those all accrue ultimately to an infrastructure opportunity that Sun's going to get after.
Describe your vision of how computing power will be consumed by the vast majority of customers five to 10 years from now.
I think you need only look to what an average consumer does to understand what the future of the enterprise will be. Most consumers use more infrastructure that's owned and operated by other people than they use their own. For example, I have my little cell phone here. I'm actually using a great deal of my operator's infrastructure to go chat with my friends and send pictures to my parents. When I go make a dinner reservation and I go to Opentable.com...I'm using their infrastructure.
The laggards in this process have been the enterprise. Some—who actually leverage Salesforce.com or Hewlett or even eBay—have figured out that the network affords them an opportunity to stop having to own and operate everything that they use. But how many consumers run their own e-mail server versus (those who) just use Yahoo Mail or Gmail? Very few. Should (companies) be using their $50 million infrastructure budget along with their management personnel, their data centers, their real estate, their power, or should they just go out and see if a buck an hour is a cheaper way of acquiring the same computer capacity?