Big Tech’s AI investment has crossed into territory that makes investors uncomfortable. Microsoft, Amazon, Meta, Google, and Apple plan to spend between $700 billion and $900 billion on AI this year.
In addition, analysts project that the figure will exceed $1 trillion in 2027. The money is moving fast and the returns are not matching these numbers.
Big Tech’s AI Investment is Triggering Investor Pushback
Investors arrived at Q1 2026 earnings asking one question above everything else, “How much more are you spending on AI?.”
First, Microsoft planned to spend 24% more than expected. Meta planned to double what it spent in 2025. Amazon committed $200 billion for the year. Alphabet also planned to spend 102% more than it did a year ago. Even Apple, known for their caution with spending, increased their AI spending.
However, markets responded immediately. Amazon fell more than 8%. Microsoft dropped over 11%. Alphabet lost 3%. But, the issue runs deeper than stock prices.
In the tech industry, over 50% of investors expect AI to start paying off within six months. Yet 84% of CEOs say that will take longer. This mismatch is causing a lot of friction, especially during earning calls.
The Strategic Logic Behind Big Tech’s AI Investment
To put it simply, companies keep spending because the alternative is worse. Tech executives believe this market rewards only the leader and under-investing now means permanent disadvantage.
In addition, Meta CEO Mark Zuckerberg cited rising memory costs as one reason spending keeps climbing. Only 6% of executives would cut AI spending even if 2026 delivers no returns.
Furthermore, 72% of CEOs now personally lead the AI investment decision, double last year’s percentage. The fear of falling behind is the only push needed.
The Infrastructure Bottleneck Slowing Things Down
Right now, infrastructure spending is causing a gap money alone can’t fix. A single AI data center now draws up to 500 megawatts of power. That rivals the electricity demand of an entire city.
Additionally, utility connections in top markets take between four and seven years to deliver. Nearly 7 gigawatts of planned U.S. data center capacity now faces delays or cancellations.
As a result, a $2 billion facility can sit idle waiting on a $40 million transformer. The constraint is physical, not financial and no amount of capital accelerates it.
The Payoff Question With No Answer
Currently, nobody can say exactly when AI will start paying off. The gap between what companies have spent and what they have earned back now stands at roughly $600 billion. Yet, the spending keeps increasing.
Ultimately, no company wants to stop spending first. Every company that keeps spending increases the pressure on the others. The risk of falling permanently behind outweighs the pressure to pause and recalibrate. This is exactly why the spending will not be slowing down anytime soon.
