As it prepares for its annual user conference in San Francisco this week, Sun Microsystems has a lot to prove. The company that once boasted of being the "dot in dot com" has watched its position as an industry leader erode over the last few years as the iconoclast image and corporate culture that served it so well during the Internet boom has worked against it during leaner times.

This tough business climate was coupled with a major management shake-up last year that saw much of Sun's senior staff exit the company, including Chief Operating Officer Ed Zander, Chief Financial Officer Michael Lehman, Executive Vice President of Computer Systems John Shoemaker, and Sun's rising Linux star, Vice President and General Manager for Edge Computing Stephen DeWitt.

Now, 14 months after its executive reorganization, the computer maker hopes to use next week's Sun Network conference to show a skeptical industry that it is still a player.

Sun has struggled to lead the industry in a variety of initiatives. Its WSCI (Web Services Choreography Interface) Web services standard has been marginalized thanks to an alternative promoted by Microsoft Corp. and IBM Corp. called BPEL (Business Process Execution Language). The Jini networking technology, created by Sun's recently departed chief scientist, Bill Joy, has failed to catch on. And Sun finds itself increasingly alone in its support of the InfiniBand I/O (input/output) architecture.

But Sun still spends US$1 billion annually on research and development, and it has high hopes for a number of developing initiatives, including its "throughput" computing microprocessor design, which insiders say will come to fruition with Sun's Niagara processors, expected in 2006; its N1 data center architecture; and with its suite of server software, code-named Orion, which will be a focal point at next week's show.

In interviews this week, Sun executives predicted that Orion, with its simplified pricing and better level of operating system integration, would change the rules of the game for the server software industry.

The question is, will the industry take notice?

Sun has always harbored a healthy contempt for conventional wisdom. Company insiders say this attitude flows down from the company's chief executive officer, Scott McNealy, who treats the term as derogatory ("If it was wisdom," he has been quoted as saying, "it wouldn't be conventional").

When competitors like IBM and Hewlett-Packard Co. adopted the Windows NT operating system in the 1990s, Sun stuck with its Solaris operating system and in the process came to dominate the Unix industry. When the 64-bit computing market was expected to consolidate around Intel's Itanium processors, Sun declined to sell Itanium systems, offering instead to port Solaris to the new chip architecture. By 2000, it dropped even this initiative, leaving its hardware competitors to develop systems based on the struggling microprocessor on their own.

But lately Sun has been following rather than leading, often in humbling fashion. When IBM decided to give away $40 million worth of software to the open source community by founding the Eclipse project in 2001, Sun officials initially dismissed the effort as a cynical effort to buy its way into the open-source tools game, and said that Eclipse's support of a nonstandard set of software libraries, called SWT (standard widget toolkit), amounted to a fork in Java.

Sun had personal reasons to be dismissive. The Eclipse project competed with Sun's own year-old open source community of NetBeans developers, and even IBM's choice of names -- with its "eclipse of the sun" connotations -- seemed to be a slight to the company that created Java.

But last week Sun confirmed that it was in negotiations to join Eclipse, a move that seemed to both confirm IBM's leadership role in Java and serve as a tacit admission that Sun's famous "write once run anywhere" vision for the programming language had not panned out, according to Clay Ryder, an analyst with The Sageza Group Inc.

"IBM has the volume to drive Java," he said. "Java is a language which has some great value to it that IBM really endorses. It's not a lifestyle like Sun sees it."

Sun has had other reversals over the last few years. It misjudged the strength of 32-bit Intel processors in its own market, killing plans to maintain a version of Solaris for x86 in January 2002, and then reversing that decision three months later. And in its highest-profile miscalculation, Sun was slow to adopt Linux; at first disparaging it, then pledging to develop its own Sun Linux distribution, and finally dropping Sun Linux in favor of partnership agreements with Red Hat Inc. and Suse Linux AG.

To a certain extent, Sun was a victim of its own success, said Larry Singer, Sun's vice president of global information systems. "We were so confident in our technology as being the best technology, we kind of took our eye off what was happening in the marketplace," he said.

The company's tendency to promote technologically exciting rather than market-tested products worked against it during the downturn, Ryder said. "When the market is growing like gangbusters , you can get away with being very forward looking and weird, but in hard times, companies fall back on the trusted and tried and true," he said.

But tried and true will get Sun only so far. The 34,000 employee company has barely managed two straight quarters of profitability, but in order to grow, Sun is betting its future on the premise that innovation still matters.

"We're not trying to outdo Dell (Inc.). That's not what we're about. That's not what our customers want," said Neil Knox, Sun's executive vice president of volume systems products. "To think that the computing world is going to go to a commodity environment where you can buy your computer from Albertson's (Inc.) or (The) Home Depot (Inc.) ... is not where the Fortune 2,000 want to go," he said.

Knox said that Sun's throughput computing chips like Niagara, with its six processing cores, will give Sun performance advantages on low cost data center computing systems that commodity vendors like Dell will be unable to achieve.

"The throughput computing chip architecture is really breakthrough and will be able to drive unmatched price performance points at the very low end," Knox said.

Sun's other big bet is that it can find a way of making money as a software company with products like Orion and the Linux-based desktop environment code-named Mad Hatter, which is expected to be publicly released next week at the Sun Network user conference.

With its per-user pricing model and a better level of integration with Solaris, Orion is the first major software release from Sun since its management shake-up a year ago. It marks the culmination of a year-long retooling of the company's software development process, which is now much more focused on integrating and coordinating software development, according to Singer.

"It's a big change in the engineering culture internally, where you have to worry about not just your products, but the integration of your products," Singer said.

"Scott (McNealy's) staff meetings used to be a competition for resources," Singer said. Now, line-of-business managers are expected to take a company-wide view in executive meetings. "From a management discipline standpoint, that's a massive transformation," Singer said.

The industry seems to agree that a massive transformation of the company is in order, but it's something that McNealy has managed before in the mid-90s when he turned a maker of technical workstations into one of the leaders of the Internet boom. McNealy has made it clear to top management that the company is going through a similar reinvention now, according to Singer.

"At first people were in denial," he said. "They've moved to a point where they say, 'OK, now this is really happening. ... The stage we're in now is the negotiation," he added. The final stage, "acceptance," is just around the corner, he predicted.