Microsoft closed out June as the most valuable wreck in the market. Its shares fell roughly 21.6% over the month, a slide that ranks as the company’s worst monthly performance since December 2000, the heart of the dot-com unwind, and one that Dow Jones Market Data flagged as potentially the steepest June drop in the stock’s history, according to Bloomberg.
The damage in dollar terms is hard to picture. The rout wiped more than $530 billion off Microsoft’s market value, a tally Bloomberg’s running headline pushed past $570 billion as the selling deepened into the close. The stock sank to one-year lows in the $349 to $353 range, ending one session at $352.83, and is now down around 24% to 25% for 2026.
Here’s the part that makes the selloff strange. Microsoft’s business is not breaking. It is, by most of the numbers a CFO would put on a slide, thriving.
A blowout quarter that investors hated
The catalyst was Microsoft’s most recent earnings report, and on the top line it was a monster. Revenue landed at $82.89 billion, up 18.3% from a year earlier. Azure, the cloud engine everyone watches, grew 40%. The company’s AI franchise crossed a $37 billion annualized revenue run rate, up 123% year over year, and Microsoft says more than 80% of the Fortune 500 now run its AI services. Those are not the figures of a company in trouble.
What spooked the market sat lower in the report. Capital spending climbed 63% from a year earlier to $38 billion in the quarter, the bill for the data centers, custom silicon, cooling, and electricity that artificial intelligence at this scale demands. Bank of America estimates Microsoft’s capital outlays for 2026 will approach $190 billion. Free cash flow — the money left over for the buybacks and dividends that reward shareholders — shrank 10% as that spending surged.
So the paradox writes itself. Microsoft is selling more AI than ever, and the cost of selling it is eating the very cash flows that justified the stock’s premium. As one framing of the problem put it, the market has stopped valuing Microsoft for what it earns today and started pricing the enormous cost of what it has to build tomorrow. That recalibration is brutal when a stock has spent three years as the cleanest expression of the AI trade.
Wall Street stops giving capex a free pass
For most of the AI boom, investors treated hyperscaler spending as a feature, not a bug: proof a company was building the future fast enough. June was the month that patience visibly thinned. Analysts began trimming their targets; Oppenheimer cut its price target on the stock to $515, citing the capex load. Measured against its peers in the S&P 500, Microsoft was among the worst performers in the index for the month, with only a small handful of companies faring worse.
None of this is unique to Redmond, which is exactly why it matters. The same questions are being asked of every company shoveling money into AI infrastructure, from the chipmakers to the cloud rivals. Microsoft has been pouring capital into the global build-out of microchips and AI data centers, and it keeps widening its partnerships, most recently bringing Anthropic’s models onto Azure to court enterprise customers. Each of those moves is a bet that the revenue eventually dwarfs the spending. The market just signaled it wants to see the revenue first.
The bull case hasn’t vanished. A business growing AI revenue at triple-digit rates with the Fortune 500 as its customer base is not a company anyone is writing off, and plenty of investors are reading the drop as a sale rather than a warning, as Quartz noted in cataloguing the one-year lows. The bear case is simpler and, for now, louder: spending is certain, payoff is a forecast, and forecasts have gotten cheaper to doubt.
Microsoft reports again this summer, and the company’s guidance on next year’s capital plan will land harder than any revenue beat. Investors spent June telling Satya Nadella’s team that the AI bill has finally grown large enough to outweigh the growth it buys. Whether that verdict holds depends on a number Microsoft hasn’t disclosed yet, the one that says when all this concrete and silicon starts paying for itself.
