As the strategic value of controlling the desktop operating system is declining, the main upsides of Google's coming launch of Chrome OS seem to be to put margin pressure on Microsoft and settling old scores.
If investing a few hundred million dollars in a branded Linux distribution with a decent browser on top can give hardware OEMs leverage in negotiations with Microsoft, Microsoft's cash flow could decrease with several hundred millions of dollars per year. A less profitable Microsoft is good for Google when the companies go head to head in capital intensive areas like data centers.
In addition tying the browser (which is really about being the default search box and to a growing extent about Apps) with an operating system gives additional distribution strength, which can actually help drive revenue.
And as Niki Scevak writes, hurting Microsoft is probably something many Google executives enjoy in itself, as quite a few of them got their butts kicked by Microsoft in the '90s. Niki:
"Eric Schmidt, a.k.a Squirrel Boy, has spent his entire career getting bitch-slapped around by Microsoft, first at Sun and then at Novell and now he is in a position to inflict damage. My problem isn't that there isn't a positive in Google pursuing a OS/Browser strategy, it's just that the majority of the effort seems to be about the negative it can inflict on Microsoft's entrenched territory in Windows. And that seems more about Eric Schmidt than Google."
Update July 15th: clarified that Microsoft's cash flow could decrease with hundreds of millions of dollars per year, not hundreds of dollars