Logica's Mike Green argues constant retirement plan buying has become the price floor under megacap tech, breaking how the market sets individual share prices and turning a correct bearish bet against AI into an untradeable bet.
Every two weeks, a few hundred dollars leaves your paycheck, lands in your 401(k), and buys a slice of the S&P 500 at the closing price. That buy does not know whether Nvidia is expensive, fairly priced, or in a bubble. It happens anyway. Mike Green argues that this passive tide has done more than buoy megacap stocks. In his reading, it has broken the mechanism that lets markets decide what a share is actually worth, and that changes what "being right" about AI even means.
The mechanism Green describes is mechanical. Index funds and target-date retirement vehicles are required to buy every constituent of the benchmark they track, in proportion to its weight, on every rebalance day. Across the roughly $13 trillion in US passive equity assets, that produces a constant bid that does not respond to fundamentals. In Green's telling, that bid acts as a price floor under the largest AI-exposed names, so anyone skeptical of Nvidia has to outgun the passive flow, not just the marginal seller. The passive-floor thesis is contested among macro researchers; it is a coherent school of thought, not consensus.
The second leg of the argument runs the other way. Levered single-stock and sector ETFs such as SOXL (3x semis) and TQQQ (3x Nasdaq) must buy or sell their underlying baskets each day to hold their stated multiple. Green estimates those rebalances run in the tens of billions daily against a much thinner passive bid for the same names. When the two flows collide, the tape can dislocate without anyone being right or wrong about Nvidia's business. Green sketches the picture in his conversation with Phil Rosen (video), and expands on the passive-flow floor in a longer interview with Excess Returns.
The Nvidia margin claim is where the vendor-financing argument enters. Green argues that Nvidia's reported gross margins reflect, in part, financing extended to customers such as CoreWeave and Nebius so those companies can afford GPU clusters. Independent coverage of the same circular-financing pattern from IO Fund lines up with Green's reading. The bull case for those margins is durable pricing power. Green's case is deferred write-down risk analogous to Cisco in 1999, when vendor financing to upstart telecom buyers turned reported router demand into later write-offs. Both sit on the same earnings line; the question is which one is the real margin and which is a loan.
That question matters because it changes what an AI drawdown would look like. If the floor under megacap tech is built from passive flows rather than from earnings conviction, an unwind could be sudden even from a price level that already reflects skepticism. The week of Green's podcast, the Korea Exchange triggered circuit-breaker halts on SK Hynix during heavy AI-linked selling, a small but visible crack at the supply-chain periphery he cites as evidence that stress has begun to surface in places the index does not directly measure. Fortune's April 2026 reporting on the gap between AI capex and hyperscaler cash returns sits in the same frame. So does Alphabet's Q1 2026 cloud beat, which shows the underlying demand is real even if the financing structure that props up the suppliers is contested.
Green's explicit positioning advice, long TIPS and the long end of the Treasury curve, is the natural trade under his thesis. Passive bond indices remain materially underweight duration, so a deflationary-AI-outcome scenario finds its expression in bonds, not in shorting megacap equity into a passive bid that does not care. That is a different object from "brace for a 1987 crash." It is a structural hedge against a market in which being right has been decoupled from making money on the trade.
The open question, and the one to watch in the next quarter, is whether vendor-financing stress shows up in earnings rather than in exchange-circuit-breaker vignettes. If it does, the same passive floor that protected megacap tech on the way up will be tested on the way down, and a correct AI bear call will finally become a tradable one.