Sun Microsystems is in deep trouble.

So say a number of analysts, most recently Brian Richardson from Meta Group who says the company has got it all wrong, with its hardware seen as proprietary and expensive.

Richardson said that it was hard to see any upside for Sun - and he's not alone. Merrill Lynch analyst Steve Milunovich wrote last month to CEO Scott McNealy and the board: "The company has gone from being pure in vision and predictable in financial performance to an underachieving, bloated, unfocused reflection of its former self."

Sun has lost money consistently ever since the end of the boom - when it boasted it was the dot in dot-com - and is about to post another set of quarterly and annual results. It declined to comment on what those results might contain.

However, revenues have declined year-on-year from $18.25 billion in 2001 to just over $11 billion in 2005. It lost $11 million in 2005.

The company has also shed thousands of jobs in recent years. In May, it announced it would slice a massive 13 percent - some 5,000 people - off its employment roster; last September it cut over 1,000 posts, which itself followed cuts of 11 percent in 2002, and 10 percent the year before.

It's lost people at the top-end of the company too, most notably last month when company co-founder Bill Joy quit, and when co-founder and chief Scott McNealy announced in May that he too was leaving.

Sun used to have a $7 billion cash pile but this year's accounts will show a sizeable dent as a result of the $4.1 billion acquisition of StorageTek which it completed last September.

While you can't attribute such decline to one single factor - execution in particular being tough to pin down - it's notable that HP and IBM have been eating Sun's lunch at the top end of the Unix market, while cheaper Linux boxes have been making serious in-roads into its systems business lower down.

In the meantime, analysts have suggested moves that Sun could make to stem the rot. They include the urgings in 2004 of Merrill Lynch to buy either Novell or Red Hat in order to get into the Linux market - a suggestion it's yet to take up.

This is hardly surprising given that Linux would offer Sun very little differentiation, and Sun trades on the perception that its products are a cut above the common herd.

On the plus side, Sun does have revenues from Java to buoy it up - a revenue stream likely to grow as more mobile phones incorporate it - plus the promise of new developments on the SPARC processor front which appears to remain its priority.

It has also open sourced Solaris and parts of UltraSPARC in a move that's been described as jumping on a bandwagon, or attempting to gain free R&D.

However, the market is consolidating on the x86 architecture, making Sun look even more like a niche player. Even Intel with massive marketing dollars has failed to capture the mass imaginations of Unix systems buyers with Itanium.

And the trend is likely to accelerate as the one-box-one-application approach becomes less important with the advent of robust virtualisation technology.

Sun has genuflected towards x86, but in a world when even Apple has abandoned the proprietary platform approach, Sun perhaps needs to rethink what it offers that's genuinely different - apart from rhetoric.