AivexaNewsSearch
AI news for builders and product teamsChecked every hour

Cracks are emerging in the AI spending boom

Collected Oct 30, 2026

Alphabet, Amazon, Meta and Microsoft have disclosed plans to spend more than $740 billion on capital expenditures this year, money Wall Street reads as AI infrastructure — data centers and chips. The reception has flipped. Where earlier rounds of spending were taken as a vote of confidence in the technology, the fourth quarter has brought a different mood: negative free cash flow, growing debt loads and doubt about whether the returns ever arrive.

That anxiety hit the market on Friday. The Nasdaq closed down 2.9%, and semiconductors sold off hard: Nvidia fell 4%, Broadcom more than 5% and Qualcomm 11%. Alphabet, which dropped 12% on Thursday in its worst trading day on record and gave up $500 billion in market value, fell again on Friday.

Alphabet is the clearest illustration of the tension. Google's parent reported negative free cash flow of $9.1 billion for the third quarter — its second straight quarter in the red — while quarterly capex rose to $51.4 billion. It expects to spend as much as $205 billion this year and has said the figure keeps climbing into 2027. At the same time, Search advertising, roughly half of revenue, grew 12%, its weakest pace in five quarters. Google's own AI Mode, the chatbot built into Search, is a runaway hit and is part of what is chipping at the ad business that funds everything else. The company is spending heavily on the technology that could erode its core franchise.

Amazon reported negative free cash flow of $42.4 billion over the trailing twelve months, against positive free cash flow of $14.8 billion a year earlier. It guides to $220 billion of capex this year. AWS revenue grew 39%, yet the stock fell 7% after Thursday's earnings. Meta posted negative free cash flow of $2.1 billion in the quarter as capex outran operating cash, with guidance up to $145 billion for the year. CEO Mark Zuckerberg has described Muse, its new free personal AI agent, as very compute-intensive to run. Ad revenue grew 26%, and Meta's shares still dropped 10% last week.

Microsoft was the outlier on cash. It recorded $18.6 billion of positive free cash flow in its latest quarter with $45.2 billion of capex, below the $50 billion it had projected. Management said that was not a deliberate retreat from AI spending but the result of capacity delays from third-party providers. Shares of smaller companies that rent out AI compute fell on the read-across.

The funding mix is shifting too. Profits are not covering the buildout on their own; equity and bond sales are filling the gap. According to S&P Global Market Intelligence, Microsoft, Alphabet, Amazon, Meta and Oracle together raised $326 billion of debt in 2026, triple the prior year's total. Demand for compute is not in question — Google, Amazon and Microsoft all posted cloud gains — but the cost of serving it is what is spooking investors.

This is a useful distinction for anyone tracking the sector: capex is a cash outflow booked against future revenue, so a quarter with negative free cash flow is not proof of a bad investment, only proof that the bill is arriving before the payoff. Depreciation on data centers and GPUs then lands on the income statement for years, which is why investors are asking how fast contracted cloud revenue converts into returns on all that hardware. The selloff suggests the market is repricing that timeline rather than rejecting AI demand outright.

The choppier backdrop could complicate Anthropic's IPO. The company filed to go public earlier this week, and its growth depends on securing very large amounts of compute. The Financial Times reported Friday that Blackstone has warned investors Anthropic might not qualify for an investment-grade rating after listing, which would leave it more dependent on the hyperscalers and third parties to finance its infrastructure.

A likely trade-off for the hyperscalers is that capital discipline becomes a public story in itself. Microsoft's shortfall against its own capex guidance moved shares of compute landlords — a sign that third-party capacity providers now trade on the spending plans of four or five customers. For model developers, the practical implication is that compute is increasingly financed with borrowed money, which tends to make contracts longer, pricing firmer and capacity less flexible when demand shifts. That is inference, not something the companies have stated.

Why it matters: Developers and product teams building on hyperscaler platforms should expect capacity and pricing decisions to be read through a financial lens, not just a technical one, as debt-funded buildouts raise the stakes on utilization. Startups dependent on rented GPU capacity are exposed to the same mood, since their suppliers' shares moved on a single capex miss. Anthropic's listing, if it proceeds without an investment-grade rating as Blackstone warned, would test how much public markets will pay for an AI lab whose growth requires heavy compute. The spending boom has not stopped; the tolerance for financing it with balance-sheet strain has narrowed.

Read at TechCrunch · AI

Based on reporting from the original publisher. Visit the source for full context and later updates.

Publisher excerpt

Big Tech is pouring billions into AI infrastructure, but mounting debt, negative cash flow, and growing investor skepticism are raising questions about the sustainability of the AI spending boom.