Online Extra: McNealy: "Computing Is Ridiculously Complex"

A: Sun is a great case study in how network computing can reduce costs. We've saved tens of millions of dollars by reducing the number of suppliers we have down to 40 and engaging in dynamic bidding with them. Our new iWork flexible-office program saved us around $50 million in real estate holdings operations. We're saving millions by using StarOffice and JavaCards throughout the company. We're a low-cost producer today. Like any CEO, I'm never satisfied. I want us to be the industry's lowest-cost producer.

Q: Some analysts argue that Sun needs to migrate to Intel chips. I know you're dead set against doing that. Why?

A: Actually, we're O.K. with it. Intel (INTC ) and AMD (AMD ) make some good 32-bit processors that we're using to go after a sweet spot in the two-way server [low-end] space, and if customers want it, we might go higher. So Sun is using x86 processors [Intel's basic chip architecture] in the parts of its product line [where it] makes sense. We also have a lot of traction with our software products, like StarOffice, that run on x86 based systems.

Our own sweet spot is in our industry-leading, 64-bit processor family [for high-end systems], which scales to 100-way systems [supercomputers] and beyond. We can get some things done with [Sun's] UltraSPARC processors that can't be done with x86 chips that give our customers real performance advantages.

In the next year, our leadership will grow as we focus on throughput -- that's application performance at a systems level. By focusing on this, we're going to turn microprocessor design on its head.

Q: How long before Intel's Itanium chip price/performance ratio catches UltraSparc in the 64-bit arena?

A: Not in our lifetimes, unless we really blow it. Don't get me wrong, Intel is a real competitor, and we watch them like a hawk. Sun is shipping its third generation of 64-bit UltraSparc systems today, while Intel's chip has been in the labs for more than a decade. We'll probably be shipping UltraSparc IV or V by the time Itanium gets any real momentum.

Q: Likewise, you're holding the line on cuts to Sun's $1.8 billion R&D budget. Why must Sun maintain its relatively high level of R&D?

A: Some of our peers agree that customers should have better solutions than they have today, that's why we all invest similar amounts in R&D. Others think everything is just fine. We just don't. Computing is ridiculously complex and costly. So do we think R&D is required to improve application performance, scalability, security, availability, manageability, and utilization? Absolutely.

Q: Do you believe Sun was slow to respond to the tech industry's downturn?

A: We hit the brakes pretty hard, but, yeah, I wish we had responded faster. My biggest regret was signing leases -- at the peak of the Bay Area real-estate boom -- for facilities we thought we would need, but didn't. On the other hand, we did a good job of monetizing the boom, putting billions in the bank.

Q: The big question for the computer industry appears to be: Is the shift toward buying cheap enterprise hardware -- or for that matter software -- a permanent shift? Or is it just a symptom of a bad economy that will go away when healthy spending returns?

A: Look, customers have always bought cheap hardware. And during a recession, IT projects shrink in size, the CFO takes over, and the customer looks to acquire smaller, less costly systems. That's why our fastest-growing products are at the low end.

Low-cost systems are here to stay, whether they're low cost at the edge of the network or low cost at the heart of the data center. We are focused on delivering the lowest acquisition and ownership costs and best throughput at every level.

There's no "one size fits all" solution to enterprise computing. You need small and large components to make all kinds of solutions. Some situations require lots of small machines working together horizontally. Others require big databases to run on big machines, and there are many situations in the middle.

What ultimately matters is the cost of running the system once you've bought it, the performance it gives you, and the choices you have when the environment changes. Are you locked in? Can the computing solution scale if business grows?

Sun has the components that allow us to simply and efficiently configure our systems to meet the needs of a broad range of requirements. We do this in a way that allows us to be the low-cost supplier while avoiding customer lock-in, yet still providing a way to change with the environment.

Overall though, I'd say technology is still too costly. It's too costly because it's too complex. Companies are spending way too much of their IT budgets on administration and maintenance, not acquisition. When we get those costs down, people will be happy to put the money they save into new servers and new storage systems. That's why we're engineering the complexity out of our systems -- so people can put more of them to work, easily and cost-effectively.

| 1 | 2 |<

Steve Ballmer, Power Forward
blog comments powered by Disqus