I left VOO last April over the inelastic market hypothesis. Here's what I found when I finally priced the AI trade I'd been avoiding.

I sold my index fund because I couldn't unsee the inelastic market hypothesis: passive flows piling into cap-weighted mega-caps regardless of price, just because that's where the next 401(k) dollar is mechanically required to go. I still believe it. I'd also be lying if I said it hasn't cost me. The index kept climbing without me, and the AI names inside it did a lot of the work. Being wrong and being early look identical for a while.

So I finally priced the thing I'd been avoiding instead of just believing my own priors. I mapped all 8 layers of the AI trade, hyperscalers, chip designers, TSMC, memory, ASML, equipment, materials, and beaten-down SaaS, and ran true FCF yield (OCF - CapEx - SBC / market cap) across roughly 30 tickers.

What I found surprised me. Adobe generates an 8.59% true FCF yield. Salesforce 7.76%. FactSet 6.10%. Nvidia, the name everyone thinks of as "the AI winner," generates 1.82%. The cash machines got sold and the dream got bought.

The section I'd most welcome pushback on: I think the "AI kills SaaS" thesis is priced in far more aggressively than the actual disruption risk justifies for names sitting on genuinely proprietary data (Adobe's 800M user behavioral dataset, FactSet's 30 years of financial model data). I don't think that's true of every SaaS name that got sold off in the same basket, just the ones with real data moats versus commodity workflow tools.

Full breakdown with the diagrams and true FCF tables: [https://cavemanscreener.substack.com/p/early-isnt-wrong-pricing-the-ai-trade](https://cavemanscreener.substack.com/p/early-isnt-wrong-pricing-the-ai-trade)