Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Good morning. On a day like yesterday, when stocks sold off, led down by tech, we found it unusual that gold, the paradigmatically defensive asset, would also fall more than 3 per cent. A delayed response to the better than expected jobs report on Wednesday? Weaker demand from China? Or has gold simply become a speculative asset?
A programming note: Unhedged will be off on Monday for Presidents’ Day. We’ll be back in your inboxes on Tuesday. Email us: unhedged@ft.com.
The software sell-off (part one)
The biggest story in stock markets in the past 15 or 20 years has been “software eating the world” — using “software” in a broad sense to include not just literal software vendors such as Microsoft but also internet services companies including Google, Meta, Amazon Web Services, as well as software-intensive hardware companies such as Apple and Nvidia. It strikes me as possible that the biggest market story in the next 15 or 20 years will be “software eating itself” — increasingly autonomous AI machines displacing not only legacy software companies but the cognitive component of all sorts of industries.
This isn’t a prediction; I don’t know enough about AI to roll out the crystal ball. But what I have read and observed leads me to think the recent sell-off in the software sector might be the leading edge of something big. Here’s a chart of the software and services sub-index of the S&P 500. It is down 27 per cent since late October:
Of course this sub-index does not capture everything that’s going on — AI panic came for trucking companies yesterday for goodness’ sake. But it’s a good microcosm to start with. And a good, if ultimately insufficient, question to start with is: is the sell-off overdone? Is there a buying opportunity in software?
Below is the valuation premium of the software and services subsector to the whole S&P 500, based on the price/earnings ratio of the two. Software is selling at a discount for only the second time in 30 years:
That’s relative valuation. The absolute valuation is not as attractive. The sector now trades at a bit above 22 times expected earnings. It was lower than that continuously from 2008 to 2016:
Of course neither relative nor absolute valuation can by itself tell you anything about whether the sell-off is overdone. For that you have to know something about what is driving the selling.
It is therefore important to recognise that we are not looking at one sell-off but two possibly overlapping ones. The first sell-off began on October 28 last year, which is when (as Unhedged has discussed at some length) there was a massive sea change in market leadership, at the expense of tech and growth stocks generally. The second started on January 28, when Microsoft’s earnings report landed and (for reasons that are not perfectly clear to me) disappointed investors badly.
Microsoft’s earnings may have been a mere catalyst for a change that was coming one way or another. But it’s still particularly important, inasmuch as it accounts for half of the software sub-index, and its 26 per cent fall since October accounts for half of the sub-index’s decline.
We can tell the two sell-offs are separate because different software stocks have performed very differently in the two periods. Here is the price return of the constituents of the sub-index in each:
Some stocks performed quite well from late October to late January, only to be crushed this month. These include Epam, which provides software engineering services and product design, and Cognizant, an information technology consultant and outsourcer. Others were crushed in the first period only to get off relatively lightly in the second. These include Oracle, a business software conglomerate and would-be cloud hyperscaler. Meanwhile, Intuit (finance and tax software), Tyler (public sector software), Gartner (tech research and consulting) and AppLovin (advertising and analytics for mobile apps) have got hit coming and going.
It is worth noting of this last group of two-time losers that they all had very wild run-ups between mid-2022 and early 2025, and were already falling when the ill wind hit software. Part of the sell-off is probably down to many members of the group getting way ahead of themselves. This chart does not include AppLovin because that company’s rise was so extreme it would make the other stocks’ gains illegible:
Scott Chronert’s strategy team at Citigroup makes a useful observation about this. Software margins had grown very wide recently and the endurance of those margins all the way to the financial horizon got priced into the stocks. That was never going to last. Chronert’s chart:
All of these comments amount to context rather than attacking the big difficult question directly. That question is whether AI machines can make legacy software companies obsolete or at the very least offer them an intense new form of competition. More on that next week.
One good read
Greenspan and Warsh and productivity.
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.
