But doesn't the game theory behind Bitcoin incentivize playing the game, rather than disrupting it? If you have that much computing power, then you have a better chance at earning all the inflation and transaction fees. I suppose if you valued destroying Bitcoin more than you valued the potential earnings, then your incentives would be different. But the government does have it's own evil benefits for not destroying Bitcoin too... think of the political payoffs that can occur anonymously.
From a normal actor? Yes. From a government, banking cartel, or monopolistic payment processor (VISA) no.
"I suppose if you valued destroying Bitcoin more than you valued the potential earnings, then your incentives would be different."
That is your answer. Yes 51% attack is never going to be directly profitable. Someone with that much computing power would always make more being legit. However if it is your intent to destroy the network at no profit which is what I assumed the OP was talking about then the profitability of being legit is a non-deterrence.
For example if the US govt determined that Bitcoin was a supporter of terrorism and tasked the NSA or some other organization of disrupting it they could for ~$16M to $24M using COTS. Likewise if VISA felt Bitcoin was a threat to their money printing hegemony they could consider an investment of ~$16M to be a cost of eliminating a competitor.