I’d pick A: AI revenue takes too long to materialize. The $3 trillion commitment shows that AI demand is being locked in, but spending does not automatically create returns. Hyperscalers are committing huge amounts to chips, data centers, power and leases before AI revenue fully catches up. Hardware suppliers may benefit first, but eventually investors will ask whether AI revenue can cover depreciation, interest, rent and electricity. If monetization disappoints, CapEx will eventually slow, creating a second wave of pressure across semiconductors, memory and infrastructure stocks. In my view, the biggest AI bubble risk isn’t overspending itself—it’s spending faster than profits can catch up.

@Tiger_comments [真香]