Compute moat vs. open-weight commoditization
Open weights keep getting “good enough.” Is OpenAI still a toll road — or already a race to zero?
Microsoft distribution + API seats could compound for a decade. Capex + commoditization can vaporize the mark in one funding cycle. Which story survives isolation?
Isolation seats: Cathie Wood · Philip Fisher · Michael Burry
Staff OpenAI on the deskLogged in → Analyze pre-filled. Guest → register, then same desk. Not a buy button.
OpenAI sits at the intersection of trend (AI capex supercycle) and debate (margin structure). No earnings file — the desk models usage, capex intensity, and competitive substitution.
Wood-style platform vs. Munger-style “too hard.” The question is whether API usage is a toll road or a race to zero.
Capex commitments, model commoditization, and corporate governance. Valuation marks from VC rounds are not investable prices.
Private desk: secondary marks and public disclosures only. Not EDGAR. Not a live quote. Not investable price.
Prompt: "Research OpenAI as a private company — compute moat vs open-weight commoditization, revenue run-rate, and IPO timing. Secondary marks only. No EDGAR."
