This article is an on-site version of our Unhedged newsletter. Premium subscribers can sign up here to get the newsletter delivered every weekday. Standard subscribers can upgrade to Premium here, or explore all FT newsletters
Good morning. It was mostly business as usual in markets yesterday: President Donald Trump wants to fire someone at the Federal Reserve, Target shares fell, and everyone was chattering about what Jay Powell would say in Jackson Hole tomorrow, especially after some hawkish-sounding minutes from the last Fed meeting. Some of the more hyped-up tech stocks did hit an air pocket, however — more on that below. As a reminder, Unhedged will be at the beach next week, but will be tanned, rested and back in your inboxes on September third. Email us: [email protected].
‘Sell-off’ non-explanations
This is what passes for a sell-off in 2025:
One popular explanation for the sell-off — if we really are insisting on calling it that — is a paper released by Nanda, an artificial intelligence initiative of the Massachusetts Institute of Technology’s Media Lab. The FT reports:
Traders pinned some of the declines in the US on a critical report on Monday authored by a branch of the Massachusetts Institute of Technology. Researchers said “95 per cent of organisations are getting zero return” from their investments in generative AI . . .
“The story is spooking people,” said one trader close to a multibillion-dollar US tech fund.
The Nanda report is an extremely silly thing to be spooked about.
“The State Of AI In Business 2025” reads like something given away on the “research” page of a large consultancy. The authors analysed a bunch of public reports about business AI implementation, sent surveys to a bunch of companies, interviewed a bunch of executives and summarised the results, adding some rather meaningless quantifications and graphs. The basic conclusions, which were notably sensible, include:
People like using out-of-the-box AI tools such as ChatGPT for grunt work, and use them even if their IT department tells them not to.
People do not like big, complicated, industry-specific systems that are hard to set up and integrate poorly with legacy technology and work processes.
A good AI system solves a small, specific problem very well and expands from there. AI systems that do a mediocre job at lots of things get binned.
A good AI system learns from its successes and failures.
Companies that try to build their own AI systems tend to mess it up.
None of this implies that AI will not be a profitable product; just that to make money, it needs to help the businesspeople it is sold to. But everyone has zeroed in on that “95 per cent get zero return” headline. Is that even a bad success rate for a new technology at this point in its history? Who knows.
The Nanda kerfuffle is a perfect example of what happens when expensive markets or sub-markets wobble a bit. After the fact, everyone looks for an explanation, and sometimes stupid explanations are the only ones available — because markets, especially expensive ones, don’t need a good reason to wobble.
Homebuilders’ shares
Yesterday we wrote about the dim outlook for housing construction in the US. What we didn’t mention then, but needs some explanation, is why homebuilders’ shares have been rallying despite the wretched backdrop:
In part, the shares are simply anticipating Fed rate cuts. John Lovallo of UBS pointed out to Unhedged that even if mortgage rates fall slightly, the positive impact on homebuilders’ margins can be significant. The builders use mortgage subsidies or “buydowns” as a sales tool, and as those buydowns get smaller, profits get bigger.
But for reasons we discussed yesterday, a lower fed funds rate might not equate to lower mortgage rates, which (mostly) vary with the 10-year Treasury rate. That said, Rick Palacios of John Burns Research and Consulting noted there is an exception to this: adjustable rate mortgages. ARMs are benchmarked to SOFR (the secured overnight financing rate), which tracks the fed funds rate. And adjustable mortgages are making a minor comeback — they now account for 8.5 per cent of all mortgages, according to the Mortgage Bankers Association. So a lower fed funds rate could help the market at the margin.
It may also be that the market has concluded things are now as bad as they can get for the housing industry — from the point of view of cost inflation, rates and consumer sentiment. With a cyclical industry, the stroke of midnight is exactly when you want to buy. But are we sure this is as dark as it gets?
More on concentration
For this week’s letter about tech’s dominance of markets, Unhedged slaved over a hot spreadsheet for hours to come up with a chart showing the value of the 10 largest US stocks as a proportion of the top 500, at five-year intervals. As it turns out, someone had done the same work, and done it better. A reader sent us this chart, from Ned Davis Research, showing the market weight of the top 10 stocks within the S&P 500 through time, going back 50 years, on a monthly basis:
We already knew that the sharp rise in concentration previous to this one, in 2000, was followed by a nasty market decline. The Ned Davis chart shows us that concentration also peaked in 1973. What the chart does not show is that the S&P 500 peaked in January of that year, and fell by almost half by September of 1974, and didn’t regain its old highs until 1980.
Does this rhyme with the dotcom bubble of 2000 and, possibly, the Mag 7/AI hype cycle of 2025? It is tempting to say so. In the late ‘60s and early ‘70s, hype was centred on the “nifty fifty” stocks which, like today’s Big Techs, were seen as one-way bets on the future. But there were a lot of other factors in the ‘73 — ‘74 crash, from the oil crisis, to the breakdown of the Bretton Woods currency regime, to Nixon’s price controls.
On a similar note, Logan Leasure of DE Shaw wrote to say the team there has looked at the market weight of the top 10 stocks within the larger market and compared it with their earnings weight over time. Here is their chart:
They argue the chart shows an important difference between 2000 and today:
In the lead-up to 2000, the weight of these companies increased notably while the earnings share stayed about the same. During the recent run-up in index weight (starting in ~2016), the earnings share of the top 10 has increased. We’d argue the two periods aren’t directly comparable, and the current period of market concentration is directionally more about fundamental economic concentration than the earlier period was.
That is right, in the sense that greater index weight is underpinned by increasing earnings share. But it is worth noting that the gap between the levels of index weight and earnings share is wide in both cases: about 10 percentage points in 2000 and about 8 percentage points now. So how you read that chart comes down to a classic question about all financial information. Do you care more about level, or the direction of change?
One good read
The new bitcoin.
FT Unhedged podcast
Can’t get enough of Unhedged? Listen to our new podcast, for a 15-minute dive into the latest markets news and financial headlines, twice a week. Catch up on past editions of the newsletter here.
Recommended newsletters for you
Due Diligence — Top stories from the world of corporate finance. Sign up here
The Lex Newsletter — Lex, our investment column, breaks down the week’s key themes, with analysis by award-winning writers. Sign up here