Online applications are no longer a trend, they're a reality. The concept of SaaS, Software as a Service, started as small drops (online e-mail) and quickly became a surge. The Online Office, also known as Office 2.0, became a ruthless battleground for giants and smaller players, eating chunks from the desktop office applications market share (see this growing list of Office 2.0 applications).
The main reason is to reduce costs. The major costs are installation and maintenance (upgrades, support). In an organization, installing and maintaining individual machines, each with somewhat different software configuration, means huge costs. This can all be simplified if the only software you'll need is your browser. An upgrade is done only on the server and most client problems can be solved by cleaning the browser cache. One may say that this is the holy grail of the IT department.
It is just a matter of time until the trend reaches software development tools. There are some examples of "online IDEs", but none of them seems strong and generic enough to replace the desktop IDE. Most of the online alternatives are focused on developing specific applications for a proprietary online platform (like Coghead). The major players in the IDE world (Microsoft, Eclipse, IBM, Sun) are still investing in desktop solutions.
Two new technologies, which has been recently introduced, suggest an online direction for Eclipse. While it's possible that none of them is adequate, this is a strong indication on the future of the platform. I don't think that the desktop version will fade away in the near future, from the same reason that desktop office applications are still with us. Nevertheless, it is possible we will see a viable online alternative to the desktop IDE soon.
The second technology is from IBM AlphaWorks, called EcliFox or "Eclipse on the Browser" (IBM are known for bad names :-). It's "An Eclipse plug-in that enables browser-based access to Eclipse". Instead of using HTML, this plug-in converts the SWT controls to XUL. This means it will only work with Gecko-based browsers, like Firefox. However, it holds the potential of running existing plug-ins, which brings us closer to the goal of having an online IDE. The demo is impressive as well.
These are very different technologies, but the goal is the same: to take the Eclipse platform on-line. I'm thinking of the amount of time spent of installing and troubleshooting the Eclipse installations in my previous project and it makes a lot of sense. This is something to keep an eye on.