We can imagine a war that causes a global blackout. So if the hash rate drops by 50%, assuming the next BTC difficulty adjustment is in 2000 blocks.
As the hash rate drops by 50%, each block will be mined in 20 minutes instead of 10 minutes. So, if the hash rate drops by 80/90/95%, it is possible that there will never be a new BTC difficulty adjustment because the time to find a nonce matching the current difficulty will increase. If my calculs are correct, for a drop of 95% of hashrate, each blocs will be mined in 200 minutes, so 277 days to reach a difficulty adjustment (considering 2000 blocs remaining).
There are an error ?
As the hash rate drops by 50%, each block will be mined in 20 minutes instead of 10 minutes. So, if the hash rate drops by 80/90/95%, it is possible that there will never be a new BTC difficulty adjustment because the time to find a nonce matching the current difficulty will increase. If my calculs are correct, for a drop of 95% of hashrate, each blocs will be mined in 200 minutes, so 277 days to reach a difficulty adjustment (considering 2000 blocs remaining).
There are an error ?
Luckily there's no such thing as a global power grid, so while large scale blackouts may happen, their effects will usually be somewhat localized (though admittedly the scale of that locale may be as large as Continental Europe or the East / West of the USA). At least if you ignore global catastrophic event like geomagnetic storms or nuclear war.
Apart from that your calculations are correct: If you cut the hashrate by 95%, average block interval increases to 200 minutes and the difficulty adjustment will take that much longer. However also keep in mind that this also assumes that none of that hashrate comes back within those 277 days. And as much as I love Bitcoin, in the unlikely case of a global blackout that knocks out 95% of the world's grid for almost a year, Bitcoin is going to be at the very bottom of my priorities
