Thanks for the very informative demonstration of the scale of these numbers. I suppose I should have emphasized this part of my post: "a weakness is found in the system that helps eliminate a huge chunk of the possible key space". I was supposing that the combined intellectual effort of the NSA and CIA could find a vulnerability in Bin Laden's crypto implementation that would reduce the size of the key space to something that could be meaningfully attacked by their combined computational capacity.
The universe contains about 2²⁶⁵ elementary particles and has existed for about 2¹⁰⁸ femtoseconds. So it probably cannot have done more than about 2³⁷³ classical computations. You can easily use triple-AES-256 to get an effective 512-bit key, dwarfing the possible computations the universe can have carried out for the foreseeable future.
The world GDP is about US$58 trillion per year, which is about 2³⁶ dollars per year. A computer that can test keys currently costs at least 2⁻⁸ dollars, although probably more, and cannot test more than about 2³⁵ keys per second, although probably less. There are about 2²⁵ seconds in a year, so that's 2⁵⁰ keys per year per processor, or 2⁹⁴ keys per year per previous year's worth of production.
So, suppose the entire world economy were devoted to producing computrons to crack a single crypto key. In N years, you can have tried ½N² · 2⁹⁴ = N² · 2⁹³ keys.
To try 2¹²⁸ keys, you need N² ≈ 2³⁵, so N ≈ 2¹⁷, a bit over a hundred millennia. (If you are willing to accept some chance of failure, say a 99.9% chance of failure, then you can skimp a bit and only try 2¹¹⁸ keys, N² ≈ 2²⁵, N ≈ 2¹², only four millennia.)
If bin Laden used AES-256 instead, you need to try 2²⁵⁶ keys to be sure of succeeding, so you need N² = 2²⁵⁶ / 2⁹³ = 2¹⁶³, so N ≈ 2⁸⁰ ≈ 10²⁴ years, which is 10¹⁴ times the current age of the universe. By comparison, the galaxies of the Local Group are expected to merge into a single supergalaxy in only 10¹² years while other galaxies are too far away to be detectable, star formation is expected to end in only 10¹⁴ years, all planetary systems are expected to have decayed in only 10¹⁵ years, and the supergalaxy is expected to have fallen apart in only 10²⁰ years. We're talking about a timescale ten thousand times longer than that. (http://en.wikipedia.org/wiki/Future_of_an_expanding_universe)
So the mere computing power of the TLAs is unlikely to yield results by brute force. Their ingenuity, however, could find a better way than brute force.