Adobe posted the first public preview of their Flash-based content runtime (codenamed) Apollo on Monday. Its pretty good - were I Microsoft, I'd be concerned.
I've discussed the ideas at some length before and Adobe's offering is clearly the strongest one out the gate... Microsoft's WPF(/e) strategy is very confused (at best), and XULRunner, from the Mozilla foundation, is potentially promising, but in practice looks to also be unsure of what its real goals are (for example, I think the ECMAScript edition 4 spec that's at its core is poorly maturing a powerful dynamic language).
But the Adobe guys seem to get what the real problems are that the browser itself solves (from a developer perspective), which is to say, a unified cross-platform development model (not for cross platform apps, per se, but to enable broadest developer knowledge) and distribution.
It's an alpha, so there's quite some goofiness, and it suffers from many of the foibles and issues that Flash does, but all in all.... its very credible as a development platform. I think the distribution and navigation aspects skew too heavily toward the desktop application paradigm, and that's a big mistake, but its one strictly of UI, not technology, so hopefully that can be addressed.
One nice bit of icing is the inclusion of a full web browsing component, enabling easy consumption of existing web content and infrastructure in your new "desktop" application. Its also the first instance of the KHTML/WebCore (the same browsing engine in Safari on the Mac and Konqeror in KDE/Linux) that's broadly available on Windows. So if you want to see how your site might look on the Mac, you can check it out with one of the Apollo samples (you can use "Scout") on Windows...
(and minor item for team Adobe: if you haven't exorcised icu and iconv from WebKit on Windows - you can save a few MB from your distribution size - by doing so...)