"This is less of a risk of a bubble, but more of a risk of lengthening monopoly gap between the S&P top ten and the remaining 490"

This is less of a risk of a bubble, but more of a risk of lengthening monopoly gap between the S&P top ten and the remaining 490. The top 10 have computational infrastructure, talent density, and data moats that compound with every AI advancement. They're not overvalued relative to their AI capability. Nope. They're accurately valued, which is the problem for everyone else.

The enterprises that matter don't compete on infrastructure anyway. Why would they? That’s why we have the hyperscalers for de-risking that element. They compete on how intelligence routes through their operational ontology, which means the real differentiation sits in the semantic architecture layer where compute becomes a commodity input.