MW 'It's existential': How Big Tech found itself in a $650 billion spending spiral
By Christine Ji
The market has spoken: It doesn't like the massive AI spending spree of the 'Big Four' hyperscalers. But the companies are barreling ahead anyway.
Can you have too much of a good thing? If the "good thing" in question is artificial-intelligence investment, then Big Tech and Wall Street are starting to disagree. And in the process, the underlying principles that have governed the AI trade thus far are eroding as well.
Alphabet $(GOOGL)$ $(GOOG)$, Amazon.com (AMZN), Meta Platforms (META) and Microsoft $(MSFT)$ - the "Big Four" hyperscalers spearheading the AI infrastructure boom - are planning to spend a collective $650 billion in 2026 to outfit data centers with chips and invest in their internal architecture. That's a 60% increase from what was spent in 2025.
A year ago, growing capital expenditures spurred stock rallies in these names, as Wall Street perceived the investment as a sign of innovation. Now, with four of the world's biggest companies gearing up to spend the equivalent of Argentina's GDP in 2026, the stakes of the AI trade are much higher. Shares of Alphabet, Amazon and Microsoft all immediately fell after their earnings announcements in recent weeks, as their spending plans overshadowed any positive news of revenue acceleration. While Meta's stock rallied initially, it rapidly gave up all of its gains in the following days.
But for the hyperscalers, AI represents both a generational opportunity and a threat to their businesses, spurring a competition to outspend the others, Michael Green, chief strategist and portfolio manager at Simplify Asset Management, told MarketWatch. AI promises to fundamentally rewire the mechanics of how the world searches, shops, advertises and writes code, leaving no corner of these tech companies' territory untouched.
"They're telling us that it's existential," Green said. "A rational market would price that. Unfortunately, we're not in a rational market."
Read: Big Tech needs a staggering $1.5 trillion to fund the AI boom. This is the complex playbook it's using to get it.
Microsoft, Meta, Alphabet and Amazon's 2026 capex plans blew past consensus expectations by roughly $150 billion. That underestimation comes perhaps as a result of Wall Street's wishful expectation that corporations should be good stewards of capital - a difficult task to accomplish if companies are spending their money on data centers instead of stock buybacks or dividends.
As the capex bill piles up, it becomes challenging to achieve a desirable return on investment. "It doesn't seem likely that the returns are going to be exceptional just because of the sheer size of the money that's being spent," John Huber, founder and portfolio manager at Saber Capital Management, told MarketWatch. To earn an "adequate" 10% return on a $1 trillion investment, the hyperscalers need to collectively generate $100 billion in incremental free cash flow every year, Huber noted.
But the opposite of that is happening. To continue funding the buildout, hyperscalers "are going from huge cash-flow generators to huge debt issuers," Green pointed out. As data centers begin to incur depreciation costs, free cash flow will only become more constrained. Amazon's plans to spend $200 billion this year could make it the first hyperscaler to experience negative free cash flow in 2026, Andrew Graham, founder and portfolio manager at Jackson Square Capital, told MarketWatch over email.
See more: This is the critical detail that could unravel the AI trade
Remaining performance obligations, or the value of future cloud deals not yet delivered, is another metric that is losing its luster amid increasing capex levels. Microsoft, Meta, Alphabet and Amazon were all capacity constrained in the fourth quarter of 2025, meaning that the demand for AI infrastructure is growing faster than they can bring data centers online.
That may seem like a promising driver of the AI boom, and it's certainly turbocharging Google and Amazon's cloud businesses, which both posted above-consensus growth for the fourth quarter.
"Obviously, there's real usage coming from model companies like OpenAI and Anthropic. ... But at the end of the chain, who are the customers?" Huber said. "For AI to to see the step function change in productivity, there has to be a benefit to the end user." To sustain the AI trade, companies will need to prove to investors that AI products can generate revenue and cash flow, Huber added.
More: Amazon's stock drops as investors question whether $200 billion can buy an AI edge
Unfortunately for investors concerned about achieving a return on investment on AI spending, it doesn't seem like there's a clear answer. When asked about details regarding Meta's revenue and return-on-investment opportunities on last week's earnings call, CEO Mark Zuckerberg responded by saying: "I think this is somewhat of an unfulfilling time to be answering some of these questions."
Huber interprets Zuckerberg's statement as a sign that nobody really knows how and when these massive investments will pay off - not even the leaders of the companies doing the spending. But the fear of missing out on AI is continuing to propel capex levels higher.
And while shareholders may be displeased with the high levels of spending, Green doesn't know if that will be enough to stop the spending, especially when the AI trade has led to massive concentration within the stock market. The "Magnificent Seven" megacap tech names currently comprise 35% of the S&P 500 SPX.
"What are investors really going to do about it?" Green said. "Sell their shares, incur massive tax costs? Step out of a market that they have made an unbelievable amount of money in for an extended period of time?"
-Christine Ji
This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
February 07, 2026 08:30 ET (13:30 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments