1. limited database access libraries, there are no standard libraries for talking to mysql/postgres/oracle/sybase/mssql
2. lack of namespace support in the language itself, which means that you can't rely on external code to not step on your variables.
Also there's a minor usability issue with using the same code in two different environments; the server side and the client-side are not the same and confusing the two can introduce whole new classes of bugs and security holes. Not that anyone here would ever lose track of where the file they were working on was supposed be run.
On 2, there are solutions to this problem based on closures, and basic politeness (YUI uses the YAHOO. prefix, while ExtJS uses Ext., and they can co-exist in one browser). It's not perfect. I'd like to see Perl-style package name spaces, or similar--nothing obtuse like the new PHP namespace support--just very concise and simple but powerful. I, personally, don't think Java style encapsulation and object border defense is all that important...but maybe enterprise users will (though Perl does OK in the stodgy banking and finance world, so it shouldn't be too scary for them).
Why?
Imagine the meeting wherein a handful of developers are tasked with choosing a language and platform for a new app. Say, four people...one likes Ruby, one likes Python, one likes PHP, and the last likes Perl (and they all agree that Java and C# are too damned verbose). Which language wins? Well, they all know JavaScript, because every web developer knows JavaScript, and they think jQuery is pretty swish...assume there's a high quality server-side environment, of course.
Obviously, it's not a foregone conclusion that JavaScript will move into a position of strength on the server. It requires too many pieces to fall into place (good implementation, good libraries, good integration with web servers, etc.) for anyone to call it a done deal. But I think it'd be foolish to bet strongly against it happening.
{I'm not baiting, I'm actually asking}
Have they? What was the biggest JavaScript application you deployed five years ago? For me, it was maybe 5k, and did nothing more than some basic form validation and content hiding effects. And now? Well, our biggest will be including almost all of ExtJS which is I think about 400k, compressed. jQuery is ~60k, compressed. Our use of JavaScript is definitely not declining. Is yours?
I think it's only in the past two years that the work of the JS library developers have proven that large JavaScript apps are plausible. They've also smoothed over most of the browser compatibility issues--which are no small part of the historic tendency to avoid doing any heavy lifting in JavaScript. Browsers are also getting better about those issues. That standardization plus quality libraries has made it more reasonable to "write once, run anywhere" with heavy JavaScript usage. Despite the cynicism with which the phrase is usually uttered, it is still a very valuable trait in a language.
I don't think AppJet has played any role in JavaScript history to date...but they might, as they're very smart fellows. Far more important players include Yahoo, Google, Microsoft, Mozilla, Adobe, and the Open Source library developers. And, of course, a few other YC alumni have been major contributors to the JavaScript juggernaut: John Resig of jQuery fame had a YC-backed startup before joining Mozilla Foundation, and Blake Ross and Joe Hewitt started YC-backed Parakey.
Anyway, it's not that I think JavaScript is perfectly suited for all tasks--or better suited for any particular task than X language. I just think it is well enough suited for most tasks (including a few for which Ruby, Python, Perl, Haskell, Lisp, etc. never will likely be at all suitable for...like the client-side of rich web apps), and the friction of switching between multiple languages, and the interoperation cost and complexity, is high enough, that JavaScript could pretty easily become the de facto standard language for web apps. And, since most of us can agree that nearly all end user apps will eventually be web apps, that means most apps.
Again, I really wouldn't want to bet against it. I'm glad I happen to find JavaScript to be a very pretty language. It has some rough spots, but overall, I don't mind working with it. It's biggest problems, I find, are caused by the DOM and its not quite comfortable connection to JavaScript.
But, if they open it up, and people agree it is awesome on a scale similar to RoR and Django, it'll grow their niche into something quite huge. They'll get a few competitors out of the bargain, but having a huge industry to fight over is better than having a tiny industry all to yourself. Heroku, even though they have far less exclusivity for their platform (deploying RoR can be done at many hosting providers), seems to be growing at a much more rapid clip than AppJet...and I believe Heroku actually launched later, but my history might be off a bit.
Anyway, I wish them luck. I'd love to see server-side JavaScript becoming more pervasive. It's a quite nice language.