An AI capital expenditure super-cycle, HBM supply falling short, and data center interconnect upgrades.

If these are the reasons you bought, none of them has been invalidated by this round of pullback. What’s being negated is only the idea that “after you buy, it should go up.”
If your position is so heavy that it affects your sleep, you need to cut—don’t hesitate. Don’t cut at the very bottom after taking losses; cut down to a level where you can confidently hold the remaining shares and wait until the end of July. A further drop of 15% after a 41% stock pullback isn’t impossible—you just need to make sure you can hold through until the catalyst arrives.
A moderate position, bought early, with a thick profit buffer. Doing nothing might be the best approach. Your profit buffer is the extra time you have compared with the market. Use that time to wait for the end-of-July earnings reports, rather than getting ground down in a slow bleed. The most expensive move during a cool-off period is frequent trading.
A moderate position, bought late, already at a paper loss. Don’t cut at this level. A cool-off period isn’t the same as hot panic—it won’t fall 50% in two days; it grinds slowly. If you cut, and then the earnings report comes out fine and the stock rebounds 15% within two days, that kind of pain is harder to bear than a paper loss. But also don’t add to the position—the reason is the same: the balance of financing hasn’t yet normalized, and deleveraging isn’t finished. Be cautious about adding. It’s not yet the so-called panic moment that brings in “bloodied chips.”
If you have no position, wait for signals. If you have a position with a profit buffer, wait for earnings reports. If you have a position without a profit buffer, don’t mess with it.