A single number on an earnings report can shatter a trillion dollars of market value in an afternoon. It’s a phenomenon that has become unnervingly common this earnings season. On paper, that number often falls under a line item investors have cheered for a decade: capital expenditures. But today, in the frenzied context of the artificial intelligence arms race, the very same spending that once signaled bold ambition is now being treated as a harbinger of doom.
I’ve long held one core belief when it comes to markets: Price is truth. Each day, a stock’s movement is trying to tell investors a story. Over time, those stories paint a fuller picture. But the harsh, almost reflexive reactions to the early batch of tech earnings are testing my patience with this belief. I fancy something is off when markets react as if building the future is a horrible thing. The story the tape is telling feels increasingly disconnected from the reality I see on the ground.
Take the case of Alphabet. The company reported second-quarter capital expenditures of $44.9 billion, a figure that sent a shiver through the market and erased roughly $293 billion in value in a single session. The full-year guidance was raised, with executives on the call pointing to a “significant” increase expected by 2027. The reason, unequivocally, is AI infrastructure. Yet, if you read the transcript, a different narrative emerges. This isn’t spending for spending’s sake. “We are seeing strong growth in Search, driven by our AI-powered overviews and planning tools,” CFO Ruth Porat noted, highlighting tangible monetization. The capital is being deployed into a revenue-generating engine that is already firing. The market’s verdict, however, was one of unvarnished panic.
This pattern repeated with brutal consistency. Tesla announced plans to commit $25 billion in capital expenditures for 2026, a tripling of its historical run rate, to ramp up its Optimus humanoid robot and robotaxi ambitions. The stock plunged 14.5%. Meta, despite stellar earnings, saw its shares wobble on capex concerns before recovering. In each instance, the chorus from the floor of the New York Stock Exchange and across trading desks was the same: the bills are coming due, and they are too high.
- Skepticism amidst capital investments
- Concerns over return on investment
- Large expenditures in AI infrastructure
- Shift in Wall Street values
- Historical commitments to growth
- A disconnect between markets and reality
This skepticism baffles me because I am hearing the exact opposite from the CEOs who are signing these checks. They aren’t nervous; they are utterly convinced. “I am absolutely convinced of the power of compute,” AMD Chair and CEO Lisa Su told me in a recent conversation. “When I say I think AI compute equates to intelligence, why wouldn’t you want more intelligence? Of course you want more intelligence.” She was quick to address the elephant in the room—return on investment. “We are seeing the return on investments. We are ramping our own AI usage within AMD very significantly month over month. And we’re seeing the productivity come back in just better products, more capable products, faster time to market.”
This sentiment is not confined to semiconductor makers. I spoke with Mastercard CEO Michael Miebach this week, who is deploying AI not as a science experiment, but as a core business tool. He’s building sophisticated fraud-prevention networks and pioneering “agentic commerce,” where AI assistants act on behalf of consumers. “It’s not a pipe dream—it’s an actual thing that is starting to happen on our platform,” he said, his tone one of palpable execution, not speculation. Even at IBM, which delivered a painful earnings warning, the story was about capital allocation to AI, not away from it. CFO Jim Kavanaugh explained to Yahoo Finance that companies are “building out the infrastructure portfolio to support that future of the AI realization.” The investment is the premise, not the problem.
So why the violent disconnect? I see two forces at play. First, there is a profound shift in what Wall Street values. For years, the market rewarded “capital-light” business models—software, platforms, networks. Heavy spending on data centers and semiconductor fabs feels like a regressive step back to the old industrial economy. It introduces depreciation, margin pressure, and execution risk that pure-play software companies often avoid. Second, there is simply the scale. The numbers are so vast they defy intuitive grasp. A $200 billion annual capex guide from Alphabet is a figure larger than the entire market capitalization of most Fortune 500 companies. The sheer weight of the commitment is staggering.
| Company | Capital Expenditure | Market Reaction |
|---|---|---|
| Alphabet | $44.9 billion | $293 billion loss |
| Tesla | $25 billion | 14.5% drop |
| Meta | Stellar earnings | Wobbled shares |
But this is where investor myopia becomes dangerous. We are witnessing the largest infrastructure build-out since the advent of the internet, perhaps since the interstate highway system. The initial outlays are enormous precisely because the opportunity is commensurate. The monetization Lisa Su and Ruth Porat describe is not hypothetical; it is occurring in real-time, in improved search yield, in enterprise software contracts, in cloud migration deals. The Federal Reserve Bank of San Francisco recently published a research note suggesting that AI investment could begin to show significant productivity gains across the broader economy within the next few years, a view echoed by the International Monetary Fund. This spending is the prerequisite for that gain.
To flee these stocks now, on the basis of these expenditures, is to make a profound bet. It is a bet that Lisa Su is wrong about the demand for compute. It is a bet that Sundar Pichai is wrong about the monetization of AI in Search and Cloud. It is a bet that Elon Musk is wrong about the value of a robotic workforce. One might disagree with any one of those visions. But to treat the collective capital allocation of the most successful technology leaders of a generation as a liability seems, to me, a fundamental misreading of the moment.
Markets are narrative machines, and right now, the dominant narrative is one of cost and fear. The alternative narrative, one of capability and future revenue, is being drowned out by the noise of the quarterly guide. My own reporting tells me the latter narrative is closer to the truth. The AI promised land is not receding; it is being constructed, beam by beam, server rack by server rack. The bills are high because the blueprint is grand. Giving these companies the cover to execute might just be the most intelligent investment thesis of all.
