Investors are pulling back on AI, but the results are coming

Figures updated to 10 November 2026.

On Thursday 5 November the Nasdaq Composite closed down 5.52%, the Nasdaq 100 down 6.36% and the Philadelphia Semiconductor Index down around 12%. In a single session the Nasdaq 100 lost around 2.5 trillion dollars in market value (AI4Business). Since then the question has been everywhere: is AI a bubble?

What investors are doing

Money is still flowing into AI, but on different terms. According to the Wall Street Journal, those who fund AI companies have become stricter and are demanding tougher conditions. Some stock market listings have been postponed and some valuations scaled back. According to the Financial Times, OpenAI cut the target of its latest fundraising round from 30 to 20 billion dollars (Sky TG24).

Some are also betting against it. According to Morgan Stanley and Goldman Sachs, for the first time since early 2023 hedge funds are holding fewer AI stocks than usual (Milano Finanza). According to the Wall Street Journal, many investors are buying put options and credit default swaps. Before the 2008 crisis, credit default swaps were used to bet against mortgages: it is the story of The Big Short. Today, though, many are using them to protect the gains of recent years rather than to bet on a crash (The Jerusalem Post, reporting the WSJ).

The reason is how much it costs to build AI: for 2026 Amazon, Alphabet, Microsoft and Meta together plan to spend around 745 billion dollars, mostly on data centres, and the market is asking how long it will take for that to pay off (LaC News24).

To give a sense of scale, Stijn Van Nieuwerburgh, professor of finance at Columbia University, has done the maths. To justify today's investment, by 2032 American businesses and consumers would need to spend around 3.5 trillion dollars a year on AI services. That would be almost 9% of the US GDP forecast for 2032: as much as they spend on food today, twice what they spend on energy (Milano Finanza).

But the results are there

Meanwhile, the big tech companies' accounts remain strong. In the third quarter the total revenues of Alphabet, Amazon, Apple, Broadcom, Meta, Microsoft and SpaceX grew by 20.5% on a year earlier, to 645 billion dollars (LaC News24).

On AI in particular, in the quarter ending 30 September Azure, the Microsoft cloud that also runs companies' AI services, grew by 48% (ANSA).

The gap shows further down, between those who invest and those who earn. A TD SYNNEX report on more than 1,500 technology resellers serving SMEs in 45 countries finds that 66% are increasing their investment in AI, but only 21% already earn a significant share of their revenue from it. And only 13% say that most of their clients' pilot projects have made it into production (BitMat).

This is how I see it: AI is delivering results, as the growth of Azure and that 21% already earning a significant share of their revenue from it show. The market is correcting how quickly it expected to make money from it.

What lies ahead

I make no predictions about the stock market. But some things are already visible.

• Fewer experiments, more measured projects. When money costs more, it goes to projects that show a result: hours saved, errors avoided, customers served better.
• The cost of every feature matters. The big providers are watching their margins too, and prices and terms can change. In one case I worked on, some automated processes kept calling the AI even with the project paused (the full case).
• Getting into production is what counts. The distance between prototypes and apps that genuinely do the work every day is, in my view, what the next few years will be about.

Those watching the stock market charts see a correction. Those who use AI to work have a chance to do the sums: what it saves them, what it costs them, what happens if something changes.

Sources

• AI4Business, 6 November 2026: «Crollo AI in Borsa: 2.500 miliardi persi, Nasdaq a -5,52%» (in Italian)
• Sky TG24, 2 November 2026: «L'AI verso una nuova bolla? Gli scenari e i timori sugli investimenti» (in Italian)
• Milano Finanza, 2 November 2026: «Gli hedge fund consigliano di sottopesare i titoli AI in portafoglio. È la prima volta dal 2023» (in Italian)
• The Jerusalem Post, 1 November 2026: «Investors turn to 'Big Short' strategies as AI stocks tumble, fears of a market crash rise»
• LaC News24, 5 November 2026: «La corsa all'AI accende i bilanci delle Big Tech, ricavi record a 645 miliardi di dollari ma Wall Street teme il conto del debito» (in Italian)
• Milano Finanza, 2 October 2026: «Gli Stati Uniti investiranno il 9% del pil nell'AI? OpenAI e Anthropic ci contano» (in Italian)
• ANSA, 29 October 2026: «La trimestrale di Google delude, Microsoft batte le attese» (in Italian)
• BitMat, 6 November 2026: «Investimenti AI in crescita, ma i ricavi non tengono il passo» (in Italian)

Do you have an app built with AI and want to know whether it is ready for production? With the AI Code Check-up I read the code and tell you what you are risking. The Full Audit also checks the costs of AI and cloud services and the integrations the app depends on. Fixed price. The AI Code Check-up page