Strip AI out of the S&P 500 and the rest has lost 7% since August, the FT's Katie Martin writes
The FT's Katie Martin says the S&P 500 without its AI component has fallen 7% since the end of August while the index itself hit a record. On her account, stocks, real estate and infrastructure are one AI position under different labels.
The Investor · Invest desk
The cap- vs equal-weighted S&P 500 gap is at its widest in nearly 24 years. The 10-year Treasury yield was above 5%, oil near $100 a barrel. Bain says AI infrastructure spending could reach $1.5 trillion a year by 2031, needing an AI market approaching $6tn a year.
- nearly 24 years Cap- vs equal-weight S&P 500 gap, widest in, claim 3
- above 5% 10-year US Treasury yield, claim 2 oil near $100 a barrel that week
- $1.5 trillion Annual AI infrastructure spending, could reach, claim 5 by 2031, Bain report cited by Martin
- approaching $6 trillion Annual AI market needed, claim 6 to sustain that investment, Bain says
| Figure | Value | Note | Claim |
|---|---|---|---|
| Cap- vs equal-weight S&P 500 gap, widest in | 24 years, nearly | 3 | |
| 10-year US Treasury yield, above | 5% | oil near $100 a barrel that week | 2 |
| Annual AI infrastructure spending, could reach | $1.5 trillion | by 2031, Bain report cited by Martin | 5 |
| Annual AI market needed, approaching | $6 trillion | to sustain that investment, Bain says | 6 |
What happened
- The records came in a week when oil held near $100 a barrel and the 10-year Treasury yield sat above 5%, either of which Martin says would usually pull stocks lower.
- Three-quarters of S&P 500 stocks fell last month, with tech and the energy names tied to the data-centre build-out holding the index up, Martin writes.
- Bain, in a recent report cited by Martin, said annual AI infrastructure spending could reach $1.5 trillion by 2031.
Why it matters
- exposure Martin writes that the AI boom now props up the dollar as well, so unhedged non-US holders of US assets carry the same trade through the currency.
- constraint Spreading money across asset classes buys less protection if they move together, and Martin says professional investors are struggling to find a way to diversify.
- contradiction Martin says US earnings justify the excitement and stocks look cheap, yet calls portfolios alarmingly reliant on one bet. A stock can be cheap on earnings and still be the same position as the others a fund holds.
The 7% is a subtraction, and the commentary does not say which stocks Martin counts as AI or what weight they carry in the index [14]. A record index sitting on a remainder that has fallen 7% needs the AI part to have gained enough to cover the loss, and the size of that gain depends on a weight we cannot see.
What the commentary does support is concentration. The gap between the cap-weighted and equal-weighted S&P 500 is at its widest in nearly 24 years [3]. Three-quarters of the index's stocks fell last month [4], so at most a quarter did not [13]. Across stocks, real estate, infrastructure, energy, emerging markets and some private equity vintages, Martin wrote, "they're all, to a large extent, the same trade" [7].
We think the concentration is better evidenced than the cause. Martin's account is that one theme is distorting reactions everywhere else [1]. The same 7% is also consistent with a 10-year yield above 5% and oil near $100 weighing on stocks outside the trade [2], and Martin herself writes that US stock prices have lagged earnings growth, "seemingly held down in part by rising bond yields" [11]. Her own definition blurs the line between the two explanations, because it puts energy stocks tied to the data-centre build-out inside the AI trade [4].
Bain's condition for the spending shows how much has to go right. According to Martin, the firm said in a recent report that "sustaining this level of investment would require an AI market approaching $6tn annually" [6]. Set against $1.5 trillion of projected annual infrastructure spend [5], that is a market four times the size of the outlay [12].
The test of the thesis is a drawdown in the AI names. If real estate, infrastructure and the ex-AI remainder fall with them, Martin is right and diversified portfolios held one position. If those assets hold, the portfolios were less concentrated than the index suggests.
What to watch
- Whether the cap-weighted and equal-weighted S&P 500 converge by the headline index falling or by the remainder rising.
- If the 10-year yield drops back below 5% and the ex-AI remainder recovers, the rates explanation gains on the crowding-out one.
- Whether non-US holders of dollar assets start hedging currency risk, which would show the unhedged exposure Martin describes being cut.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence40
- Adoption
- Insufficient
- Hype gap−10
- Incentives
- Insufficient
- Confidence50
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Katie Martin, writing for the Financial Times in a commentary republished by CNA, argues that AI has moved from being 'a' theme to 'the' theme, leaving every portfolio reliant on one bet and distorting market reactions.
- [2]
In the week the S&P 500 and Nasdaq hit records, oil held near $100 a barrel and the 10-year US Treasury yield was above 5%; Martin writes that either would usually pull stock markets lower.
- [3]
The gap between the cap-weighted S&P 500 and its equal-weighted version is not new but has reached its widest point in nearly 24 years.
- [4]
Three-quarters of the stocks in the S&P 500 fell last month; Martin writes that tech stocks and the energy stocks that reflect the data-centre build-out are doing all the work in the index.
- [5]
Private equity firm Bain said in a recent report that annual spending on AI infrastructure could reach $1.5 trillion by 2031.
- [6]
Bain added that "sustaining this level of investment would require an AI market approaching $6tn annually".
- [7]
Martin wrote of stocks, real estate, infrastructure, energy, emerging markets and some vintages of private equity: "they're all, to a large extent, the same trade."
- [8]
Non-US money has funnelled into dollar-denominated assets at an extraordinary pace over the past decade and is increasingly not hedged for currency risk; Martin writes the AI boom has become a major prop not just for the US stock market but for the dollar too.
- [9]
Martin writes that US corporate earnings, especially in tech, justify the excitement and that US stocks are pretty cheap right now.
- [10]
Martin writes that professional investors are struggling to figure out how to diversify properly and that many are spooked by how correlated portfolios turn out to be across supposedly diverse asset classes.
- [11]
Martin writes that US stock prices have not kept pace with spectacularly rapid earnings growth, "seemingly held down in part by rising bond yields".
- [12]
The $6 trillion annual AI market Bain says is needed is four times the $1.5 trillion of annual AI infrastructure spending.
- [13]
If three-quarters of S&P 500 stocks fell last month, at most 25% did not fall.
- [14]
Martin writes that if the AI component is stripped out of the S&P 500, the remaining rump has dropped 7% since the end of August, even though the index is at a record.
Get this ledger inside ChatGPT. Ask about any of our stories and the answer brings its figure, each claim’s verdict and the sources. Free, no account.
Sources
1 independent publisher whose own reporting we read for this story.
- channelnewsasia.comCommentary: Every investment portfolio looks like the same trade - AI
1 article · October 12, 2026
Topics and entities
Follow any of these and your For You feed starts watching them — no settings page required.
Topics
- AI Market ConcentrationFollow
- AI Infrastructure SpendingFollow
- US equity marketsFollow
Entities
- S&P 500Follow
- Katie MartinFollow
- CNA (Channel NewsAsia)Follow
- Bain & Co.Follow
- Financial TimesFollow