He's got the power to put Intel ahead

The boss of the mighty chip-maker tells Stephen Pritchard in Los Angeles of its plans to move into new markets and take on giants such as IBM, Cisco and Hewlett-Packard

Microchip maker Intel has achieved what very few technology companies can, and become a brand consumers actually look for. A handful of computer giants - such as Microsoft, IBM and Hewlett-Packard - might be household names, but it is the companies rather than the products that non-computer people recognise.

Perhaps only Apple, with its iPods and iMacs, can claim anything close to the success of Intel's "Intel Inside" campaign. Few of us could honestly say that we know the type of microprocessor in our machine or how it works. But we know we want it to be from Intel.

Much of this is down to a 63-year-old former researcher in materials science, Dr Craig Barrett, who joined the company as a technology development manager in 1974 and took over as Intel's president and chief executive officer from co-founder Andy Grove in 1998. It was Barrett who was responsible for the irritating logo and little jingle you hear every time anyone seems to mention the word "Intel".

Intel makes no computers of its own but sells the parts on to manufacturers such as Dell, HP and IBM. It was Barrett's decision to put the Intel logo on his customers' PCs that has done so much to ensure Intel's dominance of the market for microprocessors and other computer chips.

For desktop computers running Microsoft Windows, Intel has only one rival, AMD. In the first quarter of this year (to 15 April) Intel made profits of $915m on sales of $6.75bn. AMD, in the same period, posted a loss of $146m on a turnover of $715m.

You can't put Intel's better performance down to technology. In fact, some studies suggest that AMD's processors can outperform Intel hardware, although the computer industry is characterised by performance leap-frogging and by claims and counterclaims.

What is beyond doubt is that Intel's branding, and especially the perception that Intel is a quality brand, helps a great deal. "We have made a big investment in branding. We are a branding-driven company," says Barrett. "It serves us well." And that is from a scientist who was an associate professor at Stanford University and has published papers and even a textbook on materials science.

The problem for Intel, though, is that its chosen market, desktop computers, is maturing rapidly. Desktop sales have been little more than flat in the last couple of years, with large companies in particular cutting back on spending. And the constant, downward pressure on PC prices is cutting margins across the board. Brand or no brand, Intel is not immune.

As a result, Intel is having to find new ways to cash in on its chips. To be more specific, the company is diversifying by putting more emphasis on higher-margin servers - the central computers that run large databases or websites for companies - and on portable computing.

In servers, Intel faces far tougher competition than in the desktop computer market. Intel's success with desktops is closely linked to the popularity of Microsoft's Windows operating system, which runs on Intel or Intel-compatible chips (a fact that has not gone unnoticed by US and European antitrust regulators). The server market is more diverse, with large businesses running their software on Unix and mainframes as well as Windows.

This pitches Intel against HP, Sun and especially IBM, which makes its own processors for servers and mainframes, as well as its own operating system.

"Competition has always been there in the server industry," says Barrett. "Microsoft was slower in approaching the enterprise space than the desktop space. Enterprises have always had Unix computers. We have not seen that competition at the desktop."

Also Intel hasn't been as sure-footed in the high-end server market as it has with desktops. In a business where a single computer can cost more than $1m, buyers need compelling reasons to switch to new technologies. Intel's first attempt to take on the big Unix server market, the Itanium processor, was less than a total success.

Barrett has higher hopes for the current Itanium II processor, and the next version of this, codenamed Madison. IT managers in businesses such as banking and scientific research will often put performance before cost, and Itanium struggled to beat its rivals. Itanium II, Barrett believes, is much better.

"We said early on that that the first version [of Itanium] would find some commercial applications but its lifetime would be mostly taken up by companies porting their software to it and tuning applications," he says. "We are now at a point where an Itanium II-based HP server took the transaction processing record for Microsoft Windows Server 03. We are right there in performance and especially price-performance leadership."

Itanium II has brought Intel some significant wins. HP, which has its own version of Unix and its own processors, is now focusing on machines based on Itanium II. And Dell, previously sceptical about Itanium, will produce boxes based on the latest version.

Intel has also benefited from an unexpected ally: the open source or Linux movement. Linux is an operating system that offers much of the performance of specialist Unix systems, but at a fraction of the cost. The basic software is even available free on the internet.

But much of the cost savings of Linux come from the cheap, Intel-based hardware on which it runs. Some of the world's largest supercomputers are, in effect, networks of cheap PC servers running Linux. And IT managers like the fact that they can buy a PC server and run either Linux or Windows on it, depending on their needs.

Barrett clearly welcomes the extra business Linux brings in, but his focus appears to be more on persuading computer manufacturers such as HP or Bull, the French mainframe maker, to move to Intel chips. He has not yet given up hope that Sun - which already uses Intel processors in some of its lower-powered machines - will make the switch; he jokes that he will keep making sales calls to Scott McNealy, Sun's CEO, until he orders Itanium II.

If high-end Sun servers were to run on Intel processors, it would leave Intel with only IBM as a significant rival in the business computing market. And IBM is already an important Intel customer for its desktop computers as well as Windows and Linux servers.

"We work with a number of operating system [OS] providers to make sure their OS and applications run best on our hardware," explains Barrett. "We work with more than Microsoft. We work with [Linux] OS providers and with HP and Sun. Competition has always been there in the server industry."

Intel is also aiming at the other end of the industry: mobile computing. Laptop computers were one of the few parts of the IT industry to show real growth last year; IDC, the research group, found that sales of portables rose by 19.5 per cent in Europe while desktop sales fell 0.7 per cent. Intel itself proves the trend: of the 35,000 PCs the company is buying, 60 per cent are laptops, Barrett says.

Taking advantage of this trend, Intel launched its Centrino set of chips in March. The technology gives laptop manufacturers a one-stop shop for a power-efficient microprocessor and the electronics to connect the laptop to a wireless (Wifi) network. Wifi technology is increasingly popular both in homes and offices. And it is spreading to public places through "hotspots" in cafés, bookshops and airports.

Barrett describes Wifi as one of the few exciting developments in the IT industry today, and one, he believes, is relatively free of hype.

"The exciting thing about Wifi was that it was not a 'what if' technology," he says. "The building blocks were in place. We joined in the parade on the basis of our investment in the technology. We were able to move to the front of the parade by delivering products that do something. You could fire it up here and now. I was able to use Wifi recently myself in Brazil and Chile and Denmark."

It is by building wireless communications into laptops that Intel is taking on the networking specialists such as Cisco and 3Com. If end users prefer a Cisco wireless card, they can buy a laptop that simply runs a mobile Pentium and not the full Centrino chipset. But they might also be tempted by a computer based on an AMD processor, or the low-energy Crusoe chip from Transmeta, which Sony fits to some of its laptops.

Intel believes that the simplicity and economies of scale will persuade laptop makers to choose its solution, especially as margins come under pressure. "There will be more and more competition as the laptop segment becomes a larger part of the picture," Barrett predicts.

He is less sure, though, on the question of when the IT industry as a whole will recover. Barrett has developed something of a name for calling the bottom of the market. Intel's share price might be heading up once more, but he is now more cautious about the health of the industry.

"The IT market will get better 12 to 18 months before the communications sector," he predicts. "It will not be this month. I am still mildly optimistic that the computer part of IT will get better this year, with communications 12 to 18 months after that.

"In the enterprise sector people do understand that the cost of ownership and productivity is affected by not having the latest technology. So they have to invest."