The AI trade now wants receipts, not roadmaps, and Friday made it clear

For three years, announcing a giant AI budget was enough to make a stock go up. On Friday, that spell broke in public. Is this the start of the end of the AI boom, or just the moment the market learned to read a cash flow statement?

First, what happened. The Nasdaq closed down 2.86%, more than 8% below its record high from early October, while the S&P 500 fell 1.82% and the Dow 1.09%, according to AFP via Le Figaro. Some outlets reported the Nasdaq down nearer 3% at points during the session, but the closing figure is the one to keep. The trigger was a run of earnings: Alphabet had its worst session ever the day before (-12.64%), and then Amazon and Intel disappointed on Friday. Intel fell about 28.6%.

The contagion went well beyond those two. Nvidia lost 4.22%, AMD 10.88%, Broadcom 5.48% and Micron 5.30%. Dell dropped 10.80% and HP 9.60%. Reddit fell 29.27% after earnings in which, according to observers, the AI effect didn't show up. Even in Paris, data center suppliers such as Soitec and Schneider Electric slid. Amazon itself fell nearly 6% at one point but closed down 1.41%. The market punished it, but it didn't panic.

The market isn't punishing AI, it's punishing the bill

Look at what was actually punished. AWS grew 39%, a touch above expectations, and Microsoft's Azure surprised positively; Microsoft barely moved. Amazon still stumbled because its capital spending reached about 77 billion dollars in the quarter, against about 59 billion expected, BFM Bourse reports. Deutsche Bank's read there was that faster cloud growth wasn't enough to calm worries about rising capex.

That is the heart of the matter, in our view. Demand for AI is not the question. The question is whether the return arrives before the financing gets expensive. CEO Andy Jassy said Amazon will spend significantly more in 2027, and a Lombard Odier strategist reminded readers that a company spending more than it earns must fund the gap with debt or shares. The example he gave was Meta's capital raise, which he put at 97.5 billion dollars, the biggest in history. When a company starts selling stock to pay for its AI ambitions, the story has changed from "we are investing" to "we need your money", and investors notice.

The boom gets a sorting exercise, not a funeral

Catalyst Funds' David Miller described a widening split between companies that monetize AI effectively and those that can't justify their enormous investments. We think that is the right frame, and it is less apocalyptic than the headlines. Markets don't abandon a theme; they audit it.

Our read: the likely winners are firms where AI shows up as revenue lines you can point at, like cloud growth that outpaces spending or products customers pay for. The likely losers are those whose pitch is mostly the size of the budget. That is an opinion, not a forecast with numbers behind it, and Barclays itself says the debate over whether returns justify the outlays won't be settled any time soon. Nobody should pretend to know the answer after one ugly Friday.

Don't read every red candle as an AI verdict

Some of Friday's damage had little to do with AI. Intel's collapse came largely from PC weakness, with a global memory shortage pushing PC prices up and demand down, plus supply problems in its data center business. Hargreaves Lansdown analyst Matt Britzman said the market is in a mood of selling first and asking questions later, and that Intel's warning gave investors a convenient excuse to take profits in chip stocks. Add a Vix at its highest since April, and you have a market that was already nervous and looking for a reason.

So our advice to readers: be wary of anyone declaring the bubble popped, or the dip a gift, on the basis of one day. Watch the next earnings for two numbers: AI-related revenue growth and free cash flow. If the first outruns the second's decline, the story survives. If not, the market will keep asking for receipts, and it will be right to.