Do you have a view on sponsored surveys on total cost of ownership where the criteria are carefully chosen?

A number of surveys [are] coming out against Linux, but I don't think [they're] having any effect. I've produced things in my career where everybody has said: 'That's nice, but it doesn't really meet my requirements.' It's hard to do a study customers will take to heart because they never think their environment is like the one in the study.

Obviously, the numbers [show] people deploying Linux in a big way. It's the fastest growing operating system in the world: 25 per cent of servers and three to five per cent of desktops. The numbers don't lie.

[But] it's different around the world. In the US customers buy and deploy Linux-based solutions mostly because of lower cost, group productivity, vendor flexibility, hardware flexibility, and the software flexibility from Linux they thought they were going to get 15 years ago from Unix and the vendors took away from them.

It's completely different to Asia. Asia cares about all those things [but also] about economic development: jobs, job-creation, creating a software industry, giving people PCs for the first time running software developed in their own country by their own countrymen. That's very important to them, especially in places like India and China where people are getting PCs for the first time. [They] want a localised experience.

In Europe it's winning because of those things plus because the EU and governments around Europe are interested in open source and really behind open source initiatives. There aren't many experiments going on in the US, but there are in Europe.

Why do you think that is?

It's a combination. The [US] government takes a very literal view, but governments in Europe take a much more aggressive view. Source code matters more to governments and customers in Europe than in the US.

[In] South America, they're really sharp on price. Linux is dramatically cheaper, especially if you're not in an environment where you've ever had a PC. In the US, other than Amazon and Google, there's almost no big Linux-only company.

In the vast majority of businesses, governments and education facilities, a big chunk of servers are running Linux [but] a bigger chunk run Windows. A group of their servers run Unix [which] they're trying to figure out how to get off as fast as they can, because of the price of Risc or Unix. And new servers are going in - some Linux, some Windows.

And in fact the Windows percentage is going up, isn't it?

Yes. Because its own installed base is growing, a piece of the Unix decline goes to Windows - not a big piece - and a piece of new applications goes to Windows. It's not growing as fast as Linux, but it is growing.

Isn't it difficult to tell how fast Linux is growing because of the extra downloads?

IDC will tell you that, for every box that's shipped with Linux preinstalled they see about 0.8 Linux downloads off the internet. Then they'll tell you their research shows that, for every image shipped or downloaded off the internet, it gets copied on average 10 to 12 times because the GPL allows that.

So you see how numbers could be dramatically higher than the shipped-in boxes, which [show] 24 per cent. Twenty times the figure!

That indicates it's more than Windows on the server side.

Correct. [Then] there's a lot of $50m Linux customers who are $150m Windows customers. It's very hard for Microsoft to attack its own installed base. After Amazon and Google [Linux-only], everybody's in a mixed environment and probably has a bigger Windows bill than Linux bill. It's hard to attack your own customers, which creates a unique situation for us.

Will that tend to cause a Domino effect towards Linux?

Absolutely. Which is why, while [Microsoft] fights it as hard as it can, at some point it'll have applications to run on Linux because it can't afford to not listen to its customers.

There are people who think Linux is going to run Microsoft into the ground. I don't believe it for a minute. It's too smart, too well managed, too well run. [But] it'll listen to its customers. When it gets to some market share number, or some customer sets, or some customer rating, then some [Windows] applications will run on Linux. It'll try different things, probably in different geographies, to see what works.

Microsoft already has a 'Unix-on-Windows' environment to which it could add Linux to migrate people to Windows from Unix. But you're talking the other way round.

Customers are talking the other way round. It doesn't matter what I or Microsoft say. What's important is what the customer says. I'll say everybody's moving to Linux. It'll say nobody's moving to Linux. The fact is the data will show double-digit growth quarter over quarter. That's [what] gets reported.

What about security FUD versus reality on downtime, criticalities, vulnerabilities?

Linux runs high availability. It's very secure, running a lot of business-critical applications especially around finance [and] insurance industries. Nine out of 10 of the world's biggest supercomputers run Linux. Any time there's been an [incident] the open source community has stepped up and fixed the problem.

The good news about open source is that you don't have to wait for information to get delivered back to a company [or] for the company to mobilise people to work on the problem [or] to do the PR associated with how it's going to respond. The open source community just fixes it. The code works well and the community steps in.

There are countless examples where something broken gets fixed in Europe by the open source community before the US ever comes to work. Does that mean Linus, Andrew and the open source community do the right job of positioning that, doing the PR around that, and talking about it? No. In fact, I don't even hear about a lot because they're under no obligation to tell me.

What about the security argument that if Linux was as widely deployed then you'd have as many attacks?

Does that mean, as Microsoft grows, it's only going to get attacked more and security is going to get worse? If you say it's not as secure because more people are attacking it because of its market share, unless its market share is going to decline you have to draw a line that says it's going to be [even] less secure with more attacks and problems. That's a foolish argument.

With 25 per cent of the world's servers Linux has passed the point of critical mass. Whatever the Windows numbers are, they're both at critical mass and the security numbers speak for themselves.

How about patents and intellectual property charging?