Remember when everyone said AI was going to own the entire stock market forever?
Yeah… about that.
Thursday felt like the moment a lot of us have been quietly waiting for. The Nasdaq dropped almost 1%, Nvidia and Broadcom bled more than 3% each, and the Dow – the boring old Dow – jumped over 550 points. Money didn’t just sit on the sidelines; it actively rotated. And two very big things happened almost simultaneously to make that rotation feel real.
The Night Everything Shifted
Let me paint the picture. Wednesday night, after the bell, one of the biggest names in enterprise software reported earnings. The headline numbers weren’t terrible, but revenue came in light and – more importantly – management talked about spending even more on AI infrastructure with returns still years away. The market heard one thing loud and clear: the AI capex party is getting expensive, and the hangover might be coming sooner than the guests expected.
By Thursday morning the selling in anything labeled “AI” was relentless. Chips, cloud, software – didn’t matter. If your story was “we’re spending billions on GPUs and we’ll figure out the revenue later,” you got punished.
Then, almost on cue, the Federal Reserve wrapped up its December meeting and basically said: “Hey, the economy is actually stronger than we thought, and we think productivity could keep inflation in check even with faster growth.” Translation for traders: cyclical stocks, value stocks, finically, banks, industrials – the stuff that does well when the real economy is humming – suddenly looked attractive again.
Two Catalysts, One Big Rotation
Let’s break it down, because the timing was almost too perfect.
- Oracle’s reality check reminded everyone that hundreds of billions in AI spending doesn’t automatically become hundreds of billions in profit tomorrow.
- The Fed’s upgraded growth outlook told investors the U.S. economy might be entering a “Goldilocks 2.0” scenario – strong growth without runaway inflation thanks to productivity tailwinds.
When those two narratives collided, the result was the clearest rotation day we’ve seen in months.
It’s Not Just Oracle – It’s the Entire Spend Story
I’ve been saying this privately for a while: the market has been incredibly forgiving of companies that spend enormous amounts of cash on AI with very little revenue to show for it yet. Investors essentially gave them a multi-year hall pass because “it’s AI, it’s different this time.”
Thursday felt like the first real crack in that faith.
Think about the math. The big cloud providers and enterprises are on track to spend close to $200 billion on AI infrastructure in 2025 alone. That’s real money coming out of cash flow and balance sheets. If the revenue ramp takes longer than expected – or worse, if customers decide they don’t actually need quite as many GPUs as Wall Street modeled – then valuations built on 2027–2028 earnings hockey sticks start to look shaky.
“The catalyst fueling the rally are growing concerns about AI spend.”
— Dennis DeBusschere, 22V Research
He’s not wrong. Sometimes the thing that ends a mania isn’t a recession or a rate hike – it’s simply the realization that the emperor’s new clothes are, well, expensive.
The Fed Just Gave Permission to Own “Boring” Stocks
At the same time, the Fed basically handed value and cyclical investors a gift-wrapped excuse to rotate.
GDP forecast for 2026 got upgraded to 2.3%. Inflation expected to stay above 2% all the way until 2028. Normally that combination would terrify bond vigilantes and send rate expectations soaring. But Powell spent a lot of time talking about productivity gains potentially keeping a structural shift higher thanks to (ironically) technology.
In plain English: the Fed thinks the economy can grow faster without overheating because companies are finally getting more efficient again. That scenario is extremely friendly for banks, consumer discretionary, industrials, materials – basically everything that’s been ignored while the Magnificent Seven printed 100%+ returns.
Winners and Losers in the New Environment
Some strategists are already splitting the AI ecosystem into “winners” and “losers.” One widely followed split right now:
- Google stack → Broadcom, TTM Technologies, Celestica, Lumentum – more diversified, multiple customers, less exposure to single-point risk
- OpenAI/Microsoft stack → Nvidia, AMD, Oracle, SoftBank – heavily tied to the biggest hyperscalers’ spending plans
Thursday’s price action lined up almost perfectly with that divide.
Perhaps the most interesting part? Even if the AI spend story is intact long-term, the market is now demanding differentiation. Blind buying of anything with “AI” in the pitch deck is over – at least for now.
Where Does the Money Go Next?
If you believe the rotation has legs, the shopping list is pretty straightforward:
- Regional banks (finally benefiting from steeper yield curves and stronger loan growth)
- Consumer discretionary (especially companies tied to housing and big-ticket purchases)
- Industrials and transportation (infrastructure spend + on-shoring themes)
- Energy (still cheap on forward earnings and benefiting from power demand… including AI data centers)
- Small-caps (most sensitive to lower recession risk)
I’ve found that the best rotations happen when both the push (something wrong in the old leadership) and the pull (something clearly better somewhere else) are present at the same time. We just checked both boxes in the span of 24 hours.
Is This the Top for AI Stocks?
No. At least I don’t think so.
Long-term, the AI infrastructure build-out is probably still in the early innings. But “long-term” and “next six months” are two very different things. Valuations have sprinted far ahead of near-term fundamentals, and any hiccup in the spend-to-revenue translation was always going to be punished severely.
What we’re seeing feels more like a healthy broadening than the end of the AI trade. The 2021-2023 period where literally nothing worked outside of mega-cap tech was unnatural. Markets are cyclical. Leadership changes. Money flows to where the growth-risk-reward is most attractive.
Right now, that place appears to be almost anywhere except the most crowded AI pure-plays.
Final Thought – Watch the 10-Year Yield
If the rotation is real, the 10-year Treasury yield should continue grinding higher as growth expectations re-prices and term premium returns. That would actually be healthy for financials and terrible for long-duration tech at current multiples.
Conversely, if yields roll over again and growth fears re-emerge, the AI leaders will probably reclaim leadership in a heartbeat.
For now, though, the market is voting with its feet. And its feet are walking away from the AI trade – at least until someone shows a clearer path from hundreds of billions in spend to hundreds of billions in high-margin revenue.
Strap in. 2026 could look very different from the last three years.
