My scenario for changing the web, video production, 3D printers, games and the smart grid:
30 million for developers to unite the following into one seamless package and lobby for its inclusion into downloads of firefox/chrome/opera/safari by default (as well as being a downloadable .js for IE):
* http://jashkenas.github.com/coffee-script/ (the others would be translated into the finished language)
* http://jquery.com/
* http://nodejs.org/
* http://processingjs.org/
* http://www.highcharts.com/ (or similar with fully open licence)
Hopefully it could also be integrated into Gnome 3's javascript work and Google's Chrome OS work. Along with an opensource ogg vorbis for sound and an open source generated flash mp3 sound object the aim would be to totally replace current javascript and flash, kill silverlight, and make single client/server language development possible for those that want it. An extremely powerful "batteries included, but swappable" system on the client side (minus load times for most browsers) in addition to making node.js or sinatra-like development incredibly easy on the server side (with the potential to run micro-servers from a tab in your browser, the direction that the newest Opera technology has been heading).
The main benefit is most of the parts are already there and it could be realized very rapidly with a lot of backwards compatibility and leave IE as the "sluggish browser" getting more people to switch (or MS to implement). The main flaw would be video, but the HTML5 video tag has been discussed endlessly on HN. Lobbying could include some effort toward including ogg in more browsers and patent reform but I think 3/4 of major non-IE browsers having native theora and all having access to OS-level proprietary libraries at the same time will leave only end users to sue as NGO/community sites will just stream vorbis/theora - not ideal, but workable and better than now.
15 million on Blender 3D development split equally into three areas. Make the video editor a first-class part of the application to accelerate the opening of the pro vodcast/amateur film and documentary markets. This basically gets pro-level production down to a $800 PC and one of the new generation of digital cams (from a $150 HD up to a few thousand for a good DSLR with seperate sound recording). Second, make sure Blender is on the bleeding edge of integration with the emerging 3D printer market so that the potential there is fully realized rather than locked in (i.e. it becomes the default app for this with a good community). Third, give money to the open source game engine to promote its use in the gaming market and integration with the rest of the app (why an open source engine hasn't "done a Linux" to the underlying game infrastructure is an interesting question I think).
5 million toward an internationally certified open source electricity management hardware platform that can be manufactured by any company in the developing world for their own use and for sale to the developed world. I.e. the hardware that manages battery charge/grid interface/and is hooked up to solar and wind. A village in a developing country might have one, a guy with solar panels on his roof in a developed country might have one. Different units would be manufactured in the developing world (the design is open) and it would help economic development and accelerate the smart grid by pushing forward open standards for such devices.
Others I'd look hard at: reprap, open source cell tower hardware, opencores and arduino.