Cracks are emerging in the AI spending boom

Investors are questioning whether enormous AI infrastructure bills will eventually earn enough money. Earlier, heavy spending signaled confidence in AI. Now markets are focusing on cash leaving companies, rising debt, and the possibility that returns will disappoint.

Alphabet reported negative free cash flow of $9.1 billion in the third quarter while spending $51.4 billion on capital expenditures. Amazon reported negative free cash flow of $42.4 billion over 12 months. Meta also reported negative free cash flow. These figures show how investment can overwhelm cash generated by normal operations.

Demand for AI computing remains strong, and cloud revenue is growing at Google, Amazon, and Microsoft. Yet investors worry about the cost of supplying that demand. The concern spread across markets: the Nasdaq fell 2.9%, while Nvidia, Broadcom, Qualcomm, and major technology stocks also dropped.

Capital expenditures, or capex, are large purchases or investments in long-lasting business assets. In this article, they mainly refer to data centers, chips, and related equipment for AI computing. These investments can support future services but require cash upfront.

Free cash flow is the cash a company has left after its operating activities and capital spending. Positive free cash flow means cash generation exceeds those demands. Negative free cash flow means the company is spending more cash than it brings in. Alphabet’s third-quarter free cash flow was negative $9.1 billion, despite strong investment in AI.

AI infrastructure is the physical foundation behind AI services. Data centers house computing equipment, while chips perform the calculations. Alphabet, Amazon, Meta, and Microsoft together plan more than $740 billion in capex this year. The scale matters because infrastructure costs can weaken cash flow before new AI products produce major returns.

The four companies plan more than $740 billion in combined capital expenditures this year. Wall Street reads much of that spending as AI infrastructure investment. It includes data centers, chips, and other capacity needed to support AI services.

The individual plans show the scale. Alphabet expects to spend up to $205 billion. Amazon expects $220 billion. Meta is guiding to as much as $145 billion. Microsoft projected $50 billion, although its latest-quarter capex was $45.2 billion. Together, those stated company figures total up to $620 billion, while the article gives the broader combined estimate as more than $740 billion.

The spending is accelerating even as investors question the payoff. Alphabet expects its investment to keep rising in 2027. Amazon’s AWS revenue grew 39%, and cloud businesses at Google, Amazon, and Microsoft gained. Still, the market reacted negatively because meeting demand requires enormous cash outlays.

Revenue measures sales, not the cash remaining after expenses and investment. A company can bring in more money while paying heavily for equipment, construction, chips, workers, or other expansion. Large capital expenditures therefore can push free cash flow below zero even when the business is growing.

Amazon shows the mechanism clearly. Its AWS cloud revenue grew 39%, but Amazon reported negative free cash flow of $42.4 billion over the past 12 months. A year earlier, it had positive free cash flow of $14.8 billion. Its planned $220 billion in capex helps explain how strong business growth can coexist with cash pressure.

Negative free cash flow is not automatically proof that a company is failing. It can reflect an aggressive investment phase. But investors must judge whether future earnings will justify the spending. That question is driving concern across Big Tech, especially as AI infrastructure bills continue rising.

Big Tech’s AI investments are often funded by profitable existing businesses. The danger is that AI services may change customer behavior and reduce the value of those older businesses. Companies could then face rising infrastructure costs while their main cash engines grow more slowly.

Google provides the clearest example. AI Mode is a chatbot feature built directly into Search, but its success is eating into the economics of Google’s money-maker. Search advertising accounts for roughly half of Alphabet’s revenue. Its growth slowed to 12%, the weakest pace in five quarters, while Alphabet reported negative free cash flow of $9.1 billion.

This creates a difficult trade-off. Google must spend billions on technology that could undermine the search business paying for it. Meta faces a related cost issue: its Muse AI agent is very compute-intensive, even though advertising revenue grew 26%. Investors may therefore demand proof that AI services can replace or expand older profits.

Profits are only one funding source for AI infrastructure. Companies can also raise money by selling bonds or issuing stock. Debt provides cash now but creates repayment obligations and interest costs. Equity can provide funding without repayment, but it can dilute existing shareholders.

The article highlights debt financing. Microsoft, Alphabet, Amazon, Meta, and Oracle raised a combined $326 billion in debt in 2026, according to S&P Global Market Intelligence. That was triple the amount raised the previous year. The borrowing helps fund data centers, chips, and other infrastructure while companies continue expanding their AI capacity.

This financing adds another layer to investor concern. Companies are spending heavily, reporting weak or negative free cash flow, and borrowing more at the same time. Blackstone warned that Paradox might not qualify for an investment-grade rating after its IPO, suggesting some AI companies could become even more dependent on hyperscalers and outside funders.

AI systems require large amounts of computing capacity because they process vast quantities of data and perform many calculations. Building and running that capacity requires data centers, specialized chips, electricity, and supporting equipment. This is why AI demand translates into enormous infrastructure bills.

Cloud providers such as Google, Amazon, and Microsoft make money by offering computing capacity and related services to customers. Companies can rent that capacity instead of building all their own infrastructure. Chip companies such as Nvidia, Broadcom, and Qualcomm supply important hardware, so they can benefit when cloud providers and other businesses expand AI capacity. The article notes strong cloud growth and falling chip stocks.

Demand remains strong, but supplying it is expensive. Microsoft reported $45.2 billion in quarterly capex and cited third-party capacity delays. Shares of smaller companies that rent out AI computing power sank. The key question is whether future AI revenue will cover the infrastructure and financing costs.