So what happens if Steam goes bust tomorrow?
It won't, I can pretty much tell you that for a fact.
Will their game go bust as well since Steamworks is not redistributable? If so, it's lame to use such thing as a dependency.
Well, that's just the thing tho - Steamworks provide just about the best framework for videogame development I have ever seen. It's all in one package - reliable MP service (now bear in mind half the games I try to get running in MP just don't work, unless they use Steamworks, then they work flawlessly without any effort.
Think about it for a second, I don't think it's all that hard to imagine why would I like such a service,) handling of patches and expansion packs, effortless mod nexus feature. Now those are the major ones - then there's a lot of minor stuff, like possibility to directly tie your game with community features and make stuff like tutorial videos, walktroughs etc. easily accessible directly from your game, achievements etc.
And the last thing that's not quite as apparent to you as a customer, and that's visibility. Even if there's a lot of crap on Steam, what is and isn't on the service goes trough a redaction and selection process. Even if your title appears on the main page once, amongst new releases, it'll get a lot more hits and views than on any other service out there.
Now, can you imagine how much development time and money does all of this save? You could argue that 'But games that don't need it came out before and had such features!', but I have not seen an implementation of all these features that would be as reliable as what Steamworks provides you with. So basically, a dev studio is faced with a choice: Either they put a lot of development time and money into making these features themselves, inherently inferior to what Steam would give them freely, or they just spend more time on making the game they want to make. For many developers, that choice is really not all that difficult.
Now, you're saying that it's lame to use such a thing as a dependency - however, if open source community wants people to stop using Steam, they need to come up with a competitive API. Now that would be awesome - but chances of it happening are very small. Until there are freely redistributable dependencies with the same functionality as Steam has, decision to use it won't be 'lame', it'll be 'cost-efficient'.
If Steam does go bust, then the net code will no longer work. Same thing as if any server network goes down, or a network protocol becomes depreciated. If in 5 years we move exclusively to IPv6, will LAN play that expects a xxx.xxx.xxx.xxx address work? Or will the games lose multiplayer functionality? Sure, a wrapper may become available, but that doesn't make it ok to use a technology that may become obsolete.
I'm actually hoping that if Steam ever goes down the drain, someone manages to create a wrapper on top of its functionality. Since it's a unified API I doubt it would be that unrealistic.