For years, large U.S. technology companies appeared to have an implicit understanding with investors: they could spend heavily on artificial intelligence as long as revenue kept rising, and the stock market would continue to reward them. That arrangement is now showing signs of breaking down.

Alphabet Inc. shares fell more than 7% on Thursday, marking their worst trading day in more than a year. The drop followed the company’s decision to raise its 2026 capital expenditure outlook to as much as $205 billion. Google’s parent also reported that free cash flow turned negative in the second quarter, the first time that has happened since its 2004 initial public offering.

The market reaction came even as Alphabet reported an 82% increase in cloud-computing revenue, a result that was far above Wall Street estimates. Investors focused instead on the scale of spending.

“People are really focused on capex, obsessed with it. It used to be the more the better, but now it is the less the better,” said Jason Lemire, chief investment officer at Bold Wealth Partners. “We’re seeing capital raises, negative cash flows, rising debt. All that adds risk to the picture.”

The selloff underscored how sharply the market narrative around artificial intelligence and the Magnificent Seven technology giants has shifted. Rising capital expenditure has made it harder for companies to satisfy investors, and Alphabet’s decline was especially notable because it has been viewed as one of the largest AI beneficiaries in the group. That perception has been supported by the popularity of its Gemini AI services, its internally developed data center chips and its expanding cloud-computing business.

The pressure is centered on a basic financial trade-off. Building and running AI services requires large upfront spending on data centers, servers, networking equipment and advanced chips, while investors are looking for clearer evidence that those investments can translate into durable cash generation.

The changing backdrop creates a difficult setup for the next round of results. Microsoft Corp. and Meta Platforms are scheduled to report earnings on Wednesday, followed by Apple Inc. and Amazon.com Inc. on Thursday. Their reports will give investors another look at whether cloud growth, advertising demand and consumer hardware sales are keeping pace with the industry’s AI buildout.

An index tracking the Magnificent Seven, which also includes Nvidia Corp. and Tesla Inc., dropped 4.8% on Thursday after Alphabet’s report. It was the index’s worst day since the Trump tariff “liberation day” announcement in April 2025. The gauge is down 3.7% in 2026 after climbing sharply over the previous three years.

As a result, the companies that have dominated the S&P 500 Index since the start of the AI boom are increasingly giving up leadership to companies receiving hundreds of billions of dollars in their spending, including chipmakers such as Micron Technology Inc. and Advanced Micro Devices Inc.

Microsoft, once regarded as an AI leader because of its stake in ChatGPT owner OpenAI, is the second-weakest Magnificent Seven stock this year. Its shares have fallen 21% amid concerns that the company is losing ground despite spending more than $190 billion on capital expenditures in the current calendar year, according to analysts’ estimates. Meta shares have declined 9.8% as investors question its own AI investments, while Amazon is roughly flat for 2026.

Together, Alphabet, Microsoft, Amazon and Meta are expected to spend about $724 billion on capital expenditures this year and nearly $950 billion in 2027, according to the average of analyst estimates compiled by Bloomberg.

“We’re in a period where people are inclined to sell off on capex, and Microsoft and Meta and Amazon are all holdings hands with Alphabet and jumping in to spend,” said Willy Lee, principal at venture firm Neostellar Capital. “We’re going to see scrutiny on all parts of their businesses as they keep spending.”

The pushback from investors is also sharpening questions about the companies benefiting from the spending surge, particularly chipmakers. The Philadelphia Stock Exchange Semiconductor Index, or SOX, was up 101% through the first half of the year. It has since lost 17% in July and is on pace for its worst month since June 2022, when the stock market was in the middle of an inflation-driven selloff.

The uncertainty is visible in the recent volatility of the 30-member chip index. Its volatility over the past 100 days is at the highest level since 2020, when the pandemic was disrupting markets. The SOX has recorded 17 moves of 5% or more this year, matching the highest count since 2008, according to data compiled by Bloomberg. By comparison, the S&P 500 and the tech-heavy Nasdaq 100 Index have had none.

“There is going to be an AI winter at some point,” Lemire said. “When you look at how exceptional margins are — especially in memory — well, it is impossible to maintain those over a long timeframe. At some point, we will see margin compression and valuation compression, and that will have a huge impact on the market.”

Apple is on the other side of that trade. The iPhone maker has avoided large AI outlays, choosing instead to partner with model developers to support its services. Investors have rewarded that approach in recent weeks, sending Apple shares up 15% in July and putting them on track for their best month in exactly three years. The stock has gained 23% in 2026, making it the largest points contributor to the S&P 500’s 8.3% advance.

That does not mean Apple is without challenges. Surging demand for memory chips used in AI computing has led Apple to raise prices on products including MacBooks and iPads. How customers respond, and what the changes mean for Apple’s profit margins, remains unresolved.

The broader Magnificent Seven selloff has made some shares appear relatively inexpensive by historical measures. Microsoft, for example, trades at 19 times estimated profits, well below its average of 27 over the past decade. Meta trades at about 14 times estimated profits, compared with its 10-year average of 20.

The issue is that the rush to build AI computing capacity is changing these companies’ business models and adding new risks. Alphabet’s negative free cash flow in the second quarter drew attention because of the large amounts of cash generated by its various businesses. For companies that have long been valued partly on their ability to produce cash at scale, sustained spending can make traditional comparisons harder to interpret even when reported revenue remains strong.

That has made historical valuation comparisons less useful, according to Brad Warden, senior portfolio manager at Nomura Asset Management, whose fund holds Nvidia, Alphabet, Microsoft and Amazon.

“They look cheap right now, but when you look forward at potential disruption, they are guilty until proven innocent. Is the current business model sustainable? Will economics get worse?” said Warden, who expects the AI spenders to see returns from their investments. “It really comes down to what pain you’re willing to endure in an investment cycle and how strongly you believe you’ll ultimately get the economics on the other end of the cycle.”

This story was originally featured on Fortune.com.