AI infrastructure spending fears erase $1 trillion in market value

The cost of building AI

The world's wealthiest companies are currently engaged in a massive spending spree to build the physical foundation for artificial intelligence. According to financial disclosures, a small group of heavyweights including Amazon, Alphabet, Microsoft, Meta, and SpaceX are projected to put up to $828 billion toward capital expenses this year. This spending covers the construction of data centers, the purchase of high-powered chips, and other essential infrastructure.

The scale of this investment is reaching historic proportions, surpassing the market value of well-known corporations like Mastercard or the energy giant ExxonMobil. Some analysts at banks like Evercore and Bank of America even predict that AI-related capital expenditures could hit $1 trillion by 2027. This puts the private tech build-out on a level comparable to the roughly $1 trillion the United States spends on defense annually.

While the goal is to create the capacity needed for future AI growth, the immediate financial impact is causing friction. The sheer volume of money required to build these systems is forcing companies to move away from their traditional models of high cash reserves. Instead, they are increasingly looking to borrow money or raise new capital to keep pace with the rapid demands of the AI race.

A shift in cash flow

For years, Silicon Valley companies were known for generating massive amounts of free cash flow, which is the money left over after a company pays for its operating expenses. However, recent data shows a significant shift as these firms pour money into physical assets like buildings and equipment. Four of the five largest spenders reported negative free cash flow in the third quarter, meaning they spent more on infrastructure than they earned from their actual operations.

Microsoft was the notable exception to this trend, remaining free cash flow positive last quarter despite its heavy spending. Even so, the software giant is beginning to look for new funding sources, recently announcing a $25 billion bond offering for the first time in over nine years. This move highlights how the costs of staying competitive in the AI space are beginning to strain even the most profitable businesses.

Other giants are taking different paths to fund their expansion. Meta recently announced a plan to issue new shares to raise up to $95 billion for its AI infrastructure needs. Meanwhile, Oracle has told investors it does not anticipate a return to positive free cash flow until the latter half of its next fiscal year, which would not occur until mid-2028.

Market reactions and risks

The market's reaction to these spending levels has been sharp and immediate. On a recent Friday, stocks tumbled, causing the Nasdaq and the S&P 500 to experience their poorest single-day drops since June. During this period of volatility, Nasdaq 100 companies lost more than $1 trillion in market value as investors began to waver on the long-term viability of the AI boom.

Economists warn that these investments could have consequences that extend far beyond the walls of Silicon Valley. Because these tech giants account for an outsized share of the S&P 500, their financial health is tied to the broader economy. If these companies rely too heavily on debt to fund their expansion, it could potentially raise the cost of capital for other businesses across the country.

Aleksandar Tomic, an economist at Boston College, warned that if these investments do not pan out, the fallout could be significant. He noted that the sheer scale of the borrowing could create a situation where the rest of the market suffers if the tech leaders stumble. This creates a high-stakes environment where the success of a few companies determines the stability of the wider financial system.

Seeking a return on investment

Despite the concerns, there are signs that the demand for computing power is currently very high. Cloud providers are seeing significant growth, with Google’s cloud revenue more than doubling last quarter and Microsoft’s Azure business growing by 48%. Because cloud providers make money by renting out computing power to AI developers, these rising sales could eventually provide a welcome source of revenue for the hyperscalers.

However, rising sales do not automatically mean that the massive investments made today will be profitable in the future. Companies are currently betting huge sums of money on the capacity they expect to need years from now, which is a move that carries significant short-term risk. The central question for investors is whether AI will eventually generate enough profit to justify the enormous bill currently being paid to build it.

Amazon CEO Andy Jassy recently addressed this tension during an earnings call, acknowledging the speed of the current transition. He noted that while it might be more convenient to spread out capital expenditures over 25 years, the current opportunity requires a much faster pace. The industry is essentially in a sprint to build infrastructure, and the ultimate test will be whether that infrastructure can sustain the high costs of its own creation.