I. The consensus: get safe
The mood this week is caution about AI. On All-In, Jason Calacanis relayed Ken Griffin's view that AI is almost certainly a bubble, and that companies are asking where the productivity is on their AI bills. On the RiskReversal Pod, Cameron Dawson of NewEdge Wealth argued that risk appetite is shifting toward defensive stocks. On The Compound, a host flagged that IBM's CEO said large enterprises are scrutinizing AI spending, and that Meta just posted its first negative free cash flow quarter since 2012.
The money has moved with the mood. Health care just beat the S&P 500 by 10 points in three months. Investors are treating it as the place to wait out the AI argument.
Our view: that is a crowded place to wait. A lead this big in a defensive sector tends to shrink, and the profit growth that drives the index is still coming from the AI trade.
II. What usually happens next
We looked at every time since 2000 that health care beat the S&P 500 by 10 points or more over three months. Counting each run once, and spacing them at least six months apart, that has happened 14 times.
In the three months after, health care trailed the market 9 of those 14 times. The average result was a loss of 2.4 points against the index. The worst case gave back 13.8 points, and it is one of the most recent runs on the list. Another gave back 12.4.
The setting matters. This lead came in a rising market. Only 4 of the 14 cases looked like that, and health care faded in 3 of them. The other 10 came in falling markets, and even there the lead faded more often than it held: 6 times out of 10.
The sector is also less safe than its label. One of the biggest single-day drops among large companies this year was UnitedHealth, down 19.6% in a day. A defensive sector still holds individual stocks that can break on one quarter.
III. Where the earnings actually are
The second half of our case is about profits. Morgan Stanley Research, in a chart shared by ZeroHedge, broke down the S&P 500's 27% earnings growth for the third quarter. The Magnificent 7 supplied 26.5% of that growth. AI tech beneficiaries other than Nvidia supplied 34.9%. Everything outside AI and the Magnificent 7 supplied 38.6%.
The single biggest contributor was Micron, at 18.8% of the index's growth. Nvidia was next at 15.0%, then Meta at 6.9%, Broadcom at 5.0% and Exxon at 4.5%. None of the top five is a health care company.
The worry is about AI spending. The earnings are still coming from AI spending. When investors sell what is growing profits to buy what is steady, the steady side usually stops outperforming once the scare passes. That is what the 9 fades in 14 tries describe.
IV. The case against us
The sample is small. Fourteen episodes is a pattern, not a law. In 5 of them the lead held, and two of those held big: health care added 11.1 points in one case and 10.4 points in another. In falling markets the record is much closer to even, with the lead holding 4 times out of 10. If this is the start of an AI unwind, that weaker history is the one that applies.
There is a serious version of that unwind on the table. Gartner said "Over $3.5 Trillion of AI Infrastructure Spending Could Be Exposed to Impairment by the End of 2030," and predicted "More than 80% of neocloud providers will cease independent operations by 2030." Gartner analyst Will Sommer: "The growth in spending on AI-optimized IaaS will be too low for most AI compute providers to recover." Dawson argued the big cloud companies are earning less on each dollar they spend on AI, with returns falling from about 40% to the low-to-mid 20s, and possibly lower if they absorb smaller cloud providers' capacity.
Concentration cuts against us too. If one company supplies nearly a fifth of the index's earnings growth, a single bad forecast from Micron changes the whole picture. And this year single companies are getting repriced on one quarter at a record rate: 24 drops of 10% or more in a day among big companies through Oct 23, against an average of about 9 a year from 2010 to 2023.
If the AI earnings engine stalls, health care's lead could be the right trade and our base rate the wrong lens.
V. What settles it
Two things, both measurable.
First, the next round of earnings. If the AI names keep supplying most of the index's profit growth, the reason to hide fades, and history says the health care lead fades with it. If those profits stumble, the market likely falls, and the odds for health care's lead get closer to a coin flip.
Second, the scoreboard itself. Health care's lead as of Oct 23 is the starting line. Over the next three months, 9 of 14 past cases say it shrinks. If health care is still well ahead of the S&P 500 three months from now, this run joins the 5 that held, and that will most likely be because the AI trade broke.
The defensive trade is a bet that the bubble talk is right. The base rate says that bet usually loses in a rising market. The next quarter of AI earnings decides which history this one follows.
Disclosure: Nova One Capital may hold positions in companies we discuss. See novaonesignal.com/disclosures.
This document is for informational purposes only and does not constitute investment advice. Nova One Capital makes no representation as to the accuracy or completeness of the information herein, and past performance is not indicative of future results. October 29, 2026.