Bill Gates, Microsoft's chairman and chief software architect, came close to tagging 2006 as the release year for Longhorn, the next version of Windows, but analysts aren't so sure his implied deadline can be met.
Speaking at Gartner's Symposium/ITexpo in San Diego on Monday, Gates didn't definitively cite 2006 as the year when Longhorn will ship in final form, but said analysts' conjectures that the new operating system will debut in two years are "probably valid speculation."
But Gates warned that Longhorn's dates aren't set in stone. "Longhorn is not a date-driven release," he said. "Until you're getting close to completion, your [release] data is somewhat variable."
Gates did commit to a release of a Longhorn alpha this year. What he didn't say was anything about a beta of Longhorn, which Microsoft had earlier scheduled for this year.
The delay isn't unexpected, said Joe Wilcox, an analyst at Jupiter Research's Microsoft Monitor, even though Microsoft as recently as last May said it planned to release Longhorn in 2005. Since then, Microsoft executives have refused to commit to a definite release date for the successor to Windows XP, prompting analysts like Wilcox to theorize that its actually release date is slipped beyond even 2006.
"I'm dubious about 2006," said Wilcox. "Let's assume that Microsoft ships an alpha of Longhorn in 2004; it's a fair bet that it will be six to eight months before it hits some kind of early, or internal beta. That puts us in the second half of 2005. That would be consistent with the already announced delayed releases of Yukon and Whidbey. Those products need to be completed before serious work on Longhorn can be done."
Earlier this month, Microsoft said it would not release SQL Server 2005, formerly code-named Yukon, until the first half of next year--not during 2004, as it had previously claimed. Likewise, it rolled back the release of Visual Studio 2005, which had been labeled Whidbey, into next year.
If Microsoft manages an internal beta of Longhorn before the end of 2005, Wilcox said, Microsoft could still get Longhorn out by the end of 2006. "But that would be tough," he added.
The smart thing, he added, would be for Microsoft to run Longhorn through a long public beta, considering the amount of major changes to the operating system that it plans to introduce. Microsoft has done that before as recently as with Office 2003; during the suite's public beta testing, Microsoft extended the program by several months, delaying the product's release from early summer of 2003 to October.
Another problem with tracking Microsoft release dates, said Wilcox, is that the company's managers and executives sometimes mean public preview editions or the final release of the code--both of which can be months before actual availability--when they use the word "release."
"Apply that to the letter of the law, and Microsoft could make 2006," Wilcox said. "But realistically, for most people, Longhorn is a 2007 product."
If Microsoft wants to make good on Gates' 2006 release of Longhorn, it could conceivably shift gears in a number of ways, Wilcox said. Among the options available to Microsoft would be to ditch some of the Longhorn features Microsoft outlined last October at its developer conference. "If Microsoft begins dumping features overboard, an earlier release is possible," said Wilcox. "The classic case of Microsoft doing that was with Windows 95." When Microsoft first announced Windows 95, it said it would include features that it later dropped and rolled instead into the next versions, Windows 98 and Windows XP.
Another option would be to take the modular approach Microsoft used with Windows Server 2003--getting out a basic edition on time, then adding more components later.
But in the end, any fuss about Longhorn release dates may be moot.
"Microsoft has some maneuvering room because of its dominance," said Wilcox. "Windows is mostly the only game in town, and when that's the case, the pressure isn't as great to deliver."
And with Longhorn's release out so far into the future, it's impossible to predict what the operating system landscape may be in two or three years. "The challenge for Microsoft is what happens if Linux becomes 'good enough,'" said Wilcox. Then the pressure might be right back on.