Markets did something on Monday that is worth studying carefully, because it was not a mood. It was arithmetic.
After the chief executives of Anthropic, OpenAI and xAI publicly called for a slowdown in what they described as reckless AI development, the sell-off in anything AI-adjacent was severe. The Philadelphia semiconductor index fell almost six percent, its worst day since early July. Nvidia dropped over three percent. Broadcom and AMD fell more than four. Intel more than five, Marvell more than seven. Hewlett Packard was down ten. South Korea’s Kospi sank over three percent, hurt by a 6.4 percent fall in SK Hynix. ASML fell six in Europe.
Now the part almost nobody led with. In the same session, WPP rose 3.1 percent and the analytics group Relx rose 4.2 percent. Relx had fallen sharply earlier this year after Anthropic launched a suite of data and automation tools.
Those are the companies the market had been busy writing off. An advertising group and a professional information business are exactly the sort of firms whose revenue models have been assumed, for two years, to be on a countdown. The moment the countdown looked like it might slow, they went up.
That is the whole lesson, and it is more useful than anything in the safety debate.
A share price for a disrupted business is not a valuation of the business. It is a valuation of the time the business has left. Two things determine it: how good the business is, and how long it gets to keep being that good. The second variable is invisible, it is never disclosed, and it moves on news that has nothing to do with the company’s own performance. WPP did not get better on Monday. WPP got longer.
Every independent coffee shop within sight of a new automated kiosk understands this instinctively. When the machine goes in across the road, the little shop is not suddenly worse at making coffee. Its rent has not changed and its regulars have not left. What has changed is the assumed length of its remaining run, and that assumption governs everything downstream: whether the owner signs a five-year lease, whether they refit, whether they hire, whether the bank lends. Then the kiosk breaks down for a fortnight, or the operator pulls out of the neighbourhood, and every one of those decisions reverses, without a single cup changing hands.
The dangerous version of this is that people confuse the two variables. An owner watching the clock stops investing, and the business genuinely decays, and the decay is then cited as proof that the disruption was real. The clock became the cause. That is how a lot of businesses die, and it is entirely self-administered.
The tradeable version is what Monday showed. If you believe a firm is being priced mostly on its runway rather than its economics, then anything that extends the runway is worth more to that share price than anything the firm does. Relx up 4.2 percent on somebody else’s essay is a clean demonstration.
There is a healthy scepticism running underneath all of this that deserves a mention. Technology analysts and traders have suggested there is a messaging playbook at work, in which safety warnings invite regulation that doubles as a competitive moat. Donald Trump dismissed the calls as a sick conspiracy against AI and data centres. Not everyone thinks the slowdown is sincere, and the market will find out which it is by watching shipping schedules rather than essays.
But the price action told you something true regardless of motive.
Somebody spent Monday buying the shops that were supposed to be closing.
Source: The Guardian, “AI-linked stocks slide after tech bosses call for slowdown in ‘reckless’ development”
