2075
And likely much earlier.
Due to the nature of science and how any predictions and projections it makes needs to be couched in exceedingly conservative tones, it has become a running gag in climate science that everything will happen “much sooner than expected”. Because invariably, it does. Sometime hundreds of years sooner than expected.
Hell, it was first thought that the AMOC wouldn’t collapse for centuries, and now more accurate projections put it as being sometime between 2025 and 2085, with a “most likely due date” of some time in the early 2050s. And this is still an exceedingly conservative estimate. Who wants to bet that it’ll happen much sooner than even that?
My last reference was the 2050s “most likely due date”. That is bound to get revised radically towards the present, as more data is collected.
And at the very least, that entire range is going to be compressed towards the present as well.