
Capital is rotating out of Magnificent 7 and crypto into AI bottleneck plays — semiconductors, memory, and space stocks. The thesis is that the picks-and-shovels layer of the AI trade has more runway than the mega-cap beneficiaries already pricing in perfection.
NVDA's 65.5% revenue growth and 55.6% net margins confirm genuine earnings power behind the rotation thesis, and TSM's foundry capacity constraints mean pricing power is structural, not cyclical.
All three names have already had enormous runs and carry demanding valuations — rotation into a crowded consensus trade late in a momentum cycle historically produces sharp reversals when the next catalyst (earnings, capex update) disappoints even modestly.