carbon doesn't fatigue
Yes, it does. It's just that fatigue in carbon composites is poorly understood compared to fatigue in metals. https://link.springer.com/article/10.10 ... 8.77270.1e
though it may delaminate slowly.
Yeah, delamination is one failure mode associated with fatigue in carbon composites, especially when there are strong interlaminar shear stresses.
Interestingly, just like aluminum, carbon composites may have no fatigue limit. (The article I linked to above mentions a fatigue limit, but again, fatigue in composites is poorly understood). In orther words, there is no stress so small that it does not add to the part's accumulated fatigue. Steel and titanium both have fatigue limits, so if the stress is small enough, fatigue life is infinite. But a steel or Ti fork with infinite fatigue life would be so heavy that no one on this board would ride it. As fa63 pointed out, a reasonably well-designed part (such as an aluminum steerer) isn't going to fail in fatigue from normal use, no matter what it's made out of.
Asteroid, your anecdote is interesting, but what are you trying to say by telling it? The fact that your friend crashed (horribly!) due to a broken steerer/crown junction, while profoundly unfortunate, doesn't say anything about fatigue in either aluminum or carbon forks. It *could* have been a fatigue failure, or it could have been a manufacturing flaw. He could have overstressed the fork in a previous crash or near-crash (low-cycle fatigue), thus making the fork vulnerable to fatigue from smaller stresses (high-cycle fatigue). All forks are vulnerable to both fatigue failures and manufacturing flaws.
Detailed photos of the fracture area on both parts could hint at a fatigue failure if the break happened in metal. You typically see "shoreline" striations in the break, which shows crack growth, leading up to an irregular portion indicating brittle failure.