The chance of an "AI reckoning" has risen since mid-October. Two things moved that number up. First, Thursday and Friday wiped out tech names. Monday mattered more: the market opened with the Nasdaq higher, the S&P higher, oil down about $6, and the 10-year yield down to 4.88%. Risk assets should like lower oil and lower yields. The Nasdaq failed anyway and went lower. The one good sign is that the Nasdaq bounced off its 200-day moving average. A close below that level is needed to confirm the move continues.
Narrow market, rotation underway
Breadth was historically narrow: 35% of the S&P 500 (SPX) market cap sat in just seven stocks. Rotating out of a setup that concentrated leaves a scar. A 10% drop is not scar enough. Since the start of October a rotation has begun and is going fairly well - healthcare has led, and consumer staples have been strong. The odds are about 60/40 that the bottom drops further and the Nasdaq falls into the high teens, not just 10%.
Earnings are fine; price and positioning are the problem
The earnings have been great, continue to be great, and probably will stay great. The fundamental story does not change. What changes is market positioning and price. The lending market has pulled in huge amounts of capital. That is a linear move, and at some point someone asks whether too much money is being lent to these companies. Nvidia (NVDA) sits at the hub of this circular lending. Nvidia (NVDA) reports on November 17 (confirmed). The market has to get through that report unscathed.
The likely outcome: Nvidia (NVDA) posts stellar earnings, as it has before, and the market rewards that by selling the stock off, as it has done with most other tech firms this season. That pattern would signal a near-term top and point to a correction deeper than the 10% already seen. If Nvidia (NVDA) delivers another blowout quarter and still falls, that is a sign the AI trade is running out of buyers and steam.
A time always comes when an enthusiastic, aggressive move ends. For perspective, Micron (MU) rallied 300% in 8 weeks. A move like that is a cue to derisk. Not short these names heavily, but trim. About 80% of long tech positions have been cut, with plans to add back if the bottom drops out and a capitulation feel sets in. The AI trade is not over long term; a short-term reckoning is due, and that view is not radical - things have gotten crazy, so the hedges need trimming.
How much the economy leans on AI
AI has pulled in about $2 trillion in global investment over the last three and a half years, plus roughly $20 trillion in share price gains tied up in the AI names. If money moves out of AI, or had never been in it, that cash does not vanish from the economy - a good chunk ends up somewhere else. When the lending part of the AI trade wobbled about three months ago, people quickly went back into the 10-year Treasury and drove yields down from 5.3% to 5%. The 10-year was attractive at 5.3% and less so now near 4.95%.
That money will go somewhere else, but in the meantime the shift can hurt markets and the economy. A lot of this economy runs on more than AI project spending; it also runs on the wealth effect from stocks that have soared for three and a half years. So a tech pullback is a headwind. Countering it are several good economic forces: deregulation, different expensing rules on corporate taxes, and foreign direct investment driven by tariffs. Tariffs are not a favorite tool, but rerouting supply chains through America and the jobs that come with it is a positive.
Why a tech wobble helps the Fed
A sputtering AI growth engine would make the Fed's job easier, not harder. The Fed's position is complicated now because it is in a hiking cycle to fight inflation that is supply-side driven - and its tools do not really work on supply-side inflation. Kevin Warsh would likely welcome a reason to stop hiking if tech wobbled, because the hiking is somewhat silly.
The pattern in supply-side shocks has held for 60 to 70 years. A supply shock hits, oil and gas prices spike, and that is inflationary at first. After about a year the higher prices pull money out of people's pockets and turn recessionary. The Fed usually starts hiking about five months into the shock, and a rate hike's effect is not felt until maybe 12 months later. Those forces then converge and create a significant economic headwind. If a tech wobble knocks the stock market down significantly, it takes the pressure off Kevin Warsh and lets him sit on his hands, which is what he wants to do anyway.
The key test is the market's reaction to Nvidia (NVDA) on November 17, more than the numbers themselves, because far more than Nvidia's own results is tied to that reaction.
