There have been several rumors that a group called LargeCoin is designing a custom ASIC for hashing. If that works out, the game is over for anyone else. An array of 1000 custom ASIC chips could completely dominate the entire network. When that happens, the group that makes the ASICs can take all of the $100,000 per day of bitcoins generated and keep reinvesting it into more chips until they make the difficulty so high that no one else will ever be able to compete.
Plus ASICs will cost a small fraction of the power of any other solution. The first successful ASIC group will dominate from that point on.
Very interesting. If a single group dominated the mining computations (>50%) I would immediately stop using BTC. As far as I understand the proof of work concept they could this way simply ignore all blocks solved by others and only keep extending their 'own' blockchain version, wich would always eventually be the longest. They had total control over all transactions .. they could for exmple decide only to include transactions with a fee of X BTC. Or wouldn't it be that easy to ignore blocks found by the minor fraction of the processing power?