Wall Street Discovers Schrodinger's AI: Both Bullish and Terrified

Wall Street has entered what can only be described as a financial fever dream, where the latest earnings reports from the tech hyperscalers are simultaneously spooking bulls and bears with equal ferocity. Both sides of the market—the permabulls convinced AI will solve climate change and the permabears certain it will destroy capitalism—are finding exactly the evidence they need to validate their wildest conclusions. This is not market efficiency at work; this is mass confirmation bias wearing a suit and carrying a Bloomberg terminal.

The real trick-or-treat here is that hyperscaler earnings, by definition, contain enough mixed signals to fuel any narrative imaginable. Revenue up? Proof of AI's unstoppable march to dominance. Operating margins compressed? Evidence of excessive capex that will never generate returns. User growth accelerating? Validation. User growth decelerating? Clearly they've saturated the market and now must pivot to monetization—which is definitely different from the last pivot they announced. It's a choose-your-own-adventure earnings season where every path leads to the conclusion you arrived with.

We have seen this movie before, though the popcorn was slightly different-flavored each time. The cloud computing boom of 2010-2014 saw similar bifurcation, where every data point was elastic enough to support bull and bear cases simultaneously. The mobile arms race. The metaverse moment. The crypto supercycle. Each time, the market's collective intelligence briefly dissolved into a kind of intellectual rorschach test where two smart analysts could look at the same quarterly report and extract opposite truths.

The earnings themselves become not data points but raw material for interpretation—a Rorschach blot of balance sheet items waiting to be weaponized. When hyperscalers report that AI infrastructure investments are eating margins but driving future revenue potential, they're essentially saying: "We spent a fortune on something whose ROI is speculative, but trust us, it's good." The bulls translate this as "visionary long-term thinking." The bears translate it as "capital misallocation on a generational scale." Both are technically defensible if you squint hard enough at the financial statements and ignore inconvenient quarters.

What could go wrong? The most likely scenario is the one nobody is pricing in: AI infrastructure investments prove adequately productive, generating reasonable but not extraordinary returns, with capex discipline kicking in once deployment curves flatten. This outcome—which is the base case by historical standards—satisfies neither the bull thesis (transformative, economy-altering productivity) nor the bear thesis (catastrophic stranded assets). The market will therefore ignore it entirely and continue arguing about whether hyperscalers are genius futurists or sophisticated capital destroyers.

This earnings season has revealed that the AI trade has entered its Jekyll and Hyde era precisely because Wall Street has developed the analytical flexibility to find validation for any position. The market is not confused about hyperscalers' performance; it is perfectly organized around the principle that attention is a zero-sum game and conviction—the stronger and more contradictory, the better—drives engagement, headlines, and trading volume. In an environment where both bulls and bears can credibly claim vindication from the same quarterly numbers, it's worth asking: are we analyzing earnings, or are we just playing dress-up with our priors?

The real risk to hyperscaler stocks isn't that the market will eventually choose between bulls and bears. It's that one day, reality will arrive with its own thesis, and neither side will recognize it.