Record AI Spending Reshapes the Market's Search for 'Cash Cows' and Quality Stocks

The scale of the spending is unusual by historical standards. Columbia Business School professor Stijn Van Nieuwerburgh wrote in a paper delivered at a Brookings Institution event last week that the projected buildout would be larger relative to the U.S. economy than the country's major canal, railroad, electrification, highway and telecommunications investment booms.

That spending is pressuring cash flow at large technology companies, the report said. At the same time, the bond market is struggling to absorb the volume of AI-related debt companies want to issue, pushing more deals into junk bond territory, with AI cited among the factors behind a recent market-wide rise in yields. Some companies have pulled initial public offerings, and some deals tied directly to the AI buildout have reportedly been delayed, at least for now.

The result is a sharper focus on core balance sheet metrics, led by free cash flow, which measures cash left after expenses, interest, taxes and long-term investments. Return on equity, net debt leverage and earnings consistency have also moved higher on investors' radar, said Todd Rosenbluth, head of research and editorial at TMX VettaFi.

"With expectations that spending on AI is being revisited, investors are turning toward companies that they have confidence will continue to grow despite the shifting environment," Rosenbluth said.

Investor behavior is changing with the rate environment. During periods of low interest rates and high growth, investors have been willing to pay a premium and take more risk for future growth; with rates rising and monetary policy tightening, that appetite has cooled. One striking feature of 2026 for the S&P 500 has been a price-to-earnings ratio that has declined even as the index sits near record levels. Wall Street analysts keep raising earnings estimates, but investors are not behaving as if it is a time to speculate on valuation growth.

Shawn Snyder, economic strategist at Potomac Fund Management in Bethesda, Maryland, said investors increasingly want to avoid companies taking on excess debt, or those that could overshoot by borrowing heavily at higher rates. Most corporate debt is pegged to the 10-year Treasury, and with that yield hovering around 5%, portfolios are tilting toward higher-quality names.

"You want to focus on companies that have stronger free cash flow and are less reliant on debt at higher interest rates," Snyder said. "Cash today is worth more than the promise of cash tomorrow."

An August report by Raymond James' James Investment Group in Iowa City, Iowa, told investors the situation is bringing the sustainability of the AI sector's momentum into focus. For Zachary Evens, manager research analyst at Morningstar, investors are increasingly looking for "the legs to stand on" when it comes to AI growth, checking that companies have viable business models, a user base and a way to support the buildout before they assign value to shares.

Cash generation at the largest hyperscalers has been the anchor of the story. Until recently, those companies produced enough cash to fund investments comfortably. By the second quarter of 2026, however, aggregate capital expenditure began to exceed operating cash flow, pushing free cash flow into negative territory, the Raymond James group wrote. The firm cautioned that this is not necessarily a bearish signal: "Importantly, these companies remain extraordinarily profitable, so the issue isn't profitability. It's that AI investment has become so large that even their substantial cash flows no longer fully cover it."

Raymond James found that the technology sector of the S&P 500 had a free cash flow yield, calculated as free cash flow divided by market capitalization, of roughly 4%, about in line with the S&P 500 average. Its report concluded that free cash flow is not static and capital spending is cyclical, and that as more data centers come online, hyperscalers should be able to moderate capex while benefiting from the additional revenue those facilities generate.

The broader spending picture shows how concentrated the investment has become. Data from the U.S. Census Bureau and the St. Louis Fed cited in the report show construction spending on AI data centers has increased by $51 billion since December 2023, while private construction spending on everything else has declined by $120 billion.