It was a lament frequently heard in the 1990s and applied to many once-great systems vendors: ‘If that company fails to change direction soon, it will be the next Digital Equipment (DEC)’.

DEC was the company that made its name in the 1970s and 1980s from proprietary minicomputers – cheaper alternatives to IBM mainframes – only to miss out on the Unix wave.

Today, analysts are drawing parallels between Sun Microsystems and DEC. DEC went from being a challenger to IBM to an also-ran because it missed the ‘open systems’ revolution – a result, said observers at the time, of the obstinacy of DEC’s CEO, Ken Oslen, who once memorably condemned Unix as “snake oil”. Sun, as a pioneer of Unix, was arguably the company that signed DEC’s death warrant.

But as Sun continues to falter, critics are jumping on the comparison. For Ken Olsen, read Sun’s CEO, Scott McNealy. And for Unix, read Linux – the increasingly popular open source operating system that Sun first ignored and then eventually supported, albeit tepidly.

That is the backdrop against which Sun released its results for the opening quarter of the company’s fiscal 2004. While rivals IBM and Unisys posted strong sales growth and profits for the same period, Sun’s revenues sank 8% to $2.54 billion and net losses increased, from $111 million in the same quarter a year ago to $286 million. No one is yet writing Sun’s obituary, but the disappointing results have provoked some harsh comment from analysts.

Chief among those urging radical measures is Steve Milunovich, the influential analyst at investment bank Merrill Lynch analyst. He supported Sun’s R&D-led strategy during the IT boom, but he now says that a change in direction is needed. He recently sent an open letter to Sun executives urging them to cut at least 7,000 staff, embrace commodity technology (including using Intel processors in preference to Sun’s own Sparc chips), ditch the recently launched ‘Mad Hatter’ desktop system and spin off Java technology.

Sun has dismissed such suggestions as ill-informed. But there is no doubt that Sun’s costs are higher than many of its peers. For example, the company spends 16% of its revenue on R&D, compared to just 1% at Dell, 6% at IBM and 6% at Hewlett-Packard (HP). And its selling, general and administrative (SG&A) expenses amount to 29%, far higher than at IBM and HP.

In any case, the unexpected departure of co-founder and chief scientist Bill Joy indicates possible strategic changes at Sun may be in the offing. As a pointer to that, recent reports in Japan have suggested that Sun and Japanese systems giant Fujitsu are considering merging their Unix systems development.

Perhaps one positive aspect of Sun’s recent results is the slow, but steady increase in the percentage of revenues derived from services – up from 28% in the first quarter of fiscal 2002, to 36% in the latest quarter.

Given that, Unisys might provide a more reassuring blueprint for Sun than DEC. Like DEC, Unisys struggled for much of the 1990s. But whereas DEC eventually succumbed to a $9 billion takeover from Compaq, Unisys ditched its PC and low-end server businesses, pushed deeper into services and increasingly concentrated its hardware development on the niche, but highly profitable, market for high-end Microsoft Windows multi-processor servers.

And that appears to have paid off. In the third quarter to the end of September, Unisys posted a healthy revenue increase of 9% to $1.45 billion with revenue from services growing even more strongly, up by 11% to $1.12 billion and now representing 77% of total sales.

However, that figure also includes a significant amount of maintenance revenues related to Unisys’ legacy mainframe business that still draws on its 1986 creation from the merger of Sperry and Burroughs.

Nevertheless, Unisys CEO Larry Weinbach also reported a 50% increase in sales of its high-end ES7000 Windows servers, based on symmetrical multiprocessor (SMP) clustering technology, a performance helped along by a reseller pact with Dell.

However successful Unisys has been in shifting the epicentre of its business from hardware to services, it is still dwarfed by IBM and its Global Services arm. That division is now the most profitable part of the systems giant, according to chief financial officer John Joyce.

IBM Global Services’ results for the three months ending 30 September were inflated by the addition of PwC Consulting which IBM acquired in July 2002. This helped push revenues at Global Services up by 17% to $10.4 billion, and resulted in the services unit accounting for 48% of IBM’s total revenues in the quarter of $21.5 billion.

Meanwhile, the company’s revenue backlog – the payments that are due to flow from long-standing outsourcing contracts but which are yet to be recognise on IBM’s books – also rose for the first time in recent quarters, up from $112 billion to $115 billion.

However, analysts see some less buoyant times ahead. Annex Research analyst Bob Djurdjevic says Global Services is under price pressure as a result of clients renegotiating deals that were struck at the height of the technology boom. That suggests the services business is looking at a period of slower growth and thinner margins.