Kudos on announcing it. Now we're going to help hold you accountable! :)
1. Amazon Web Services:
What parts of Amazon Web Services are you planning on using? AWS S3 is great for image/file storage and sharing, and dirt cheap too. Hook it up behind AWS Cloudfront and you get a nice fast content distribution network.
The crummy part of AWS is EC2 - poor performance compared with real VPSs like Linode or Digital Ocean. (Just google "EC2 vs Linode"). What are you hosting on currently that EC2 would be an extreme speed increase? I guess if one is using AWS' database services too, then EC2 would make sense. Just be careful - committing fully to the AWS ecosystem could be much more costly than going with something more generic, especially if you become dissatisfied with it later on and want to move away from it.
2. Underground Academy
Similar to the Manifesto, these videos could be something that draws people in initially. Some videos could be for members only, while some are open to the public. Like cgcookie.com, a video site which focuses on the more technical/art side of game dev, sometimes the first video in a series is open to the public, while the rest of the series are for members only or cost a membership fee.
Also, I'd love to see more useful interviews with industry people - ask successful indie devs, marketers, publishers, review sites, etc - they have different takes on things. Kind of like that one GDU chat night with the guys from IGM.
3. Teams and project management tools
The "project management tools built in to help you and your team communicate and work together" - that sounds big in scope. I don't think that should be prioritized. What value would that tool need to add, that devs can't already get much better value from sites that specialize in project management & team communication such as Trello?
Maybe shift the focus to tools that help the game dev process. Examples:
- Game Design Document maker - interactive questions (branching tree of questions, text fields, and multiple choices) that help a scattered game dev create and flesh out their design doc, and then spits out an initial GDD (text-based or html). The question tree might branch based on genre/features. Has useful tips and "have you considered this and that..." for most question.
- Marketing plan wizard - handholdy tool that takes you from zero marketing plan to a full marketing plan, asking you questions, making you do tasks, and spits out the customized step by step marketing plan, timeline, and checklist.
- Game hook evaluator/augmenter - a bunch of questions that are rough predictors of how successful a game hook/plan is, answer and self grade (or have others grade), to see what its score is, and generate you recommendations on how to augment your game (See the book "David Perry on Game Design A Brainstorming Toolbox" - tons of great questions and evaluations)
- Matchmaking/barter board: there are a lot of solo developers who may need to someone with the skills they don't already have. This helps devs to form teams or get access to skills. Perhaps they can post what skills they need, and what they're willing to pay/trade for it. (ex: Unity expert needs concept artist. Willing to trade for C# programming time or testing time, hour for hour.)
- Member map - maybe you need to find some new teammates locally for collaboration/hands-on-testing, a Google map with pins of members' locations (based on their self-entered location field, not the NSA GPS snooping type) would be useful. Could filter this by "last active" so you can find only the people active on GDU within the last X weeks.
Thought experiment: what would happen to GDU if it is opened up (no invites needed, anyone can join), or only some parts are invite-only? Pros & cons, new opportunities, opportunities lost. What value does it add to keep GDU invite-only?
This may be private, but how is this going to be funded? Performance hosting takes money. With invite-only, it would not be anywhere near enough traffic to generate the hosting cost through ads. The system takes lots of time and/or money to build AND maintain AND enhance/update so that it doesn't go stale.