Expanding on its utility computing offering, Sun Microsystems earlier this week unveiled an online storage service and launched a website where consumers can buy computing cycles at a rate of $1 per CPU hour. Prior to the official launch, vnunet.com spoke exclusively with Sun's president and chief operating officer, Jonathan Schwartz.

What is 2005 going to bring?
The market will grow. The internet will see continuing growth, not only of users but of devices. The value of the network will increase. The vision we have espoused for 23 years of 'The network is the computer' will be true once again.

All these things suggest that bandwidth is like electricity and water: a commodity that spans the planet and for which there is a global demand. In terms of what people do with bandwidth, that's as predictable as what people do with electricity or water.

Last year you started talking about $1 per CPU cycle, but we haven't heard much since.
We haven't started talking about the customer wins. What we have seen is a large number of CIOs who are now benchmarking their data centres and trying to figure out if they are spending more than a buck an hour. I see a huge amount of proofs of concepts where customers are looking at what they are paying for their own grid or what they are paying outsourcers.

They are figuring out that they are overpaying by a factor 10 to 100. The migration towards technology as a service is a cultural shift. The number of people who use search today versus the group that used search 10 years ago, is an unbelievable multiple because Google made it simple. But moving to Google is a cultural shift, not a technical one.

Most businesses today buy computers. My belief is that most businesses five years from now will be buying computing. It's not that one will make the other go away. It's just that there is a dominant model that says it's better to buy electricity from the wall than make your own with your own generator that you run with your own staff and your own chief electricity officer.

Speaking of Google, when is it going to switch to Solaris 10?
When it stops creating its own private version of its own private operating system and stops building its own custom version of its own custom hardware. Google is hiring mechanical engineers and I guess most of my corporate customers don't want to be doing system board layout and fan design. They want to be running systems that other people are providing. Google is a little bit of an anomaly.

Isn't it weird that one of the most scalable networks in the world - Google's search network - doesn't see the vision that Sun is trying to put out there?
No. I find it more interesting that more people haven't created services on the web like Google has. I don't know a single bank that is creating its own database like Google is, or its own application system like Google is.

I don't think Google is necessarily the norm to which I expect to see a lot of the corporate customers migrating. Google is an anomaly that takes advantage of customising technologies for a very explicit business purpose.

What questions should CIOs be asking themselves today?
What are the cultural shifts that I need to drive that will allow me to get to spending a fifth as much on IT as I do today? Most challenges in IT today are not technical; most of the challenges are cultural. Whether it's leveraging $1 per CPU hour or a $100 per employee for middleware or leveraging a Linux-based desktop or moving towards open source, those are cultural changes, not technical changes.

Why is it so hard to change that culture?
Because culture is far harder to evolve than technology. We can try to make it safe; to make sure you can point out beacons or highlight customers that are saving a lot of money. But ultimately culture changes over generations. It doesn't necessarily change in an instant.

Even the internet is only just beginning to come into its own now. And it was introduced back in 1994. It took a decade to get it adopted and fully endorsed and deployed.