Got to get to work but I would point out SHA-2 (commonly called SHA-256, SHA-384, and SHA-512) has not been broken, not even theoretically.
An attack on a reduced round version of SHA-2 has been found what this means is that SHA-2 uses 64 rounds (the input is passed through the same algorithm 64 times). IF SHA-2 only used 42 rounds a faster than brute force attack would be possible. However I would point out that the time complexity of such an attack is 2^251. A brute force preimage attack on SHA-2 requires 2^255 operations so the "attack" is 2^(255-251) = 2^4 = 16 times more faster than brute force. That still requires more time and energy than our solar system has to achieve a collision.
Lastly SHA-2 being partially broken (i.e. much faster than brute force attack) is not a concern when it comes to mining. If miners using a modifed algorithm implementing the attack could mine 10,000 faster then it would simply mean difficulty would rise by a factor of 10,000. If SHA-2 is broken completely then the issue isn't mining it is the privacy of the public key however LTC (and alll? alt-coins) use the same address structure. This is less of a threat as addresses are the double SHA-256 and RIPEMD-160 hash of the public key. It is unlikely both algorithms would be broken sufficiently to provide a credible threat.
The largest cryptographic threat to BTC (and by extension all alt-coins as they have copied this portion of the code bit for bit) is ECDSA. Public key cryptographic is much harder to predict the likelihood that the cipher will remain secure. Unlike hashing algorithms public key cryptography relies on an "unknown". The strength comes from the fact that there is currently no feasible method of solving certain math problems. If that assumption turns out to be false then the cipher will be vulnerable. Through either cryptoanalysis or quantum computing I full expect ECDSA (and many other public key systems) to be broken wide open within my lifetime. An alt-coin using an alternative public key system would provide an "insurance" policy of sorts but AFAIK none exist.
A cryptographic system by definition is theoretically broken if an attack faster than brute force is known.
Which doesn't exist for SHA-2 because it uses 64 rounds. The attack was on a 42 round variant not used by anyone. The attack doesn't work (even theoretically) against the actual SHA-2.
Up went the diff on BTC to the point GPU miners could not make a profit. Now since GPU miners make up about 80%+ of the miners what did you think they were gonna do, sit by while the old guard has steak to eat, while they get bread crumbs. They are moving to the scrypt coin and nothing can be done about it. BTC are only worth something because people say they are. Well the 80%+ is getting ready to say BTC is NOT worth anything anymore. Hence LTC is rising.
Lets assume all this is true (which it isn't). LTC uses a weakened version of scrypt one which ASICs can be produced cost effectively so what happens when BTC "dies" and LTC become mainstream. The global mining revenue shoots up into the hundreds of millions of dollars a year which justifies the development of LTC ASICs?
Despite the scrypt memory requirement relaxation in LTC we are still talking about 4 or 5 orders of magnitude of higher cost. Nobody came ever close to an efficient LTC ASIC even in theoretical designs, so the point is the development of LTC ASICs is by orders of magnitude less justified than the development of BTC ASICs and therefore won't happen before LTC gains in price as much.
5 orders of magnitude really?
So BTC ASIC costs say $1M NRE and $5 per chip.
LTS $10B NRE and $500,000 per chip.
Really. I mean that is just wrong that is common sense stupidly wrong.
An LTC ASIC can be produced today however the tiny amount of global mining revenue doesn't warrant it yet. The efficiency gains are lower but the costs are not so high as to be a barrier. Bitcoin ASICs didn't start looking cost effective until the exchange rate (and thus potential global mining revenue) was high enough.