Big Tech’s June-quarter earnings landed within days of each other, and together they tell a clear story about AI infrastructure spending — and a messier one about what the market is willing to pay for it.

Microsoft set the tone: revenue of $90.01 billion, up 18%, and EPS of $4.74 both cleared estimates, with Azure growing 43% and crossing $100 billion annualized. Shares jumped 8% after hours.

Amazon was arguably the standout. Revenue topped $200 billion for the first time, up 20%, while AWS grew 36.7% to $42.2 billion, its fastest pace since late 2021. Backlog hit $496 billion, and capex guidance rose to $220 billion from $200 billion. The stock still rallied double digits.

Meta broke the pattern. Revenue of $60.8 billion beat, but EPS of $6.18 missed on legal and severance charges, and quarterly capex of $31.08 billion left free cash flow at just $784 million. Shares fell roughly 7%.

Apple beat and got sold anyway. Revenue of $109.4 billion rose 16%, iPhone 22%, Mac 29%, but Services disappointed and gross margin fell 120 basis points sequentially to 48.1%, a drop management attributed almost entirely to memory costs. September-quarter growth guidance of 9-11% reflects supply constraints on advanced nodes.

For semiconductors, the message is that demand is settled. Combined 2026 capex across Microsoft, Amazon, Meta, Apple, and Alphabet now sits well above $600 billion, and Apple says it’s production-limited by leading-edge wafer availability. Every one of those data points is bullish for chipmakers.

Yet chip stocks fell into a bear market in July, down more than 20% from June highs. The market has stopped paying for demand and started scrutinizing how it’s financed: Amazon’s trailing free cash flow is negative $7.6 billion, and Alphabet went cash-flow negative for the first time since its IPO. Capex funded by debt behaves differently in a downturn than capex funded by operations.

A reasonable framework for the next 12 months: favor leading-edge foundry and logic, where Apple’s constraint actually binds; treat memory as a trading position rather than a hold, given the volatility; stage entries around catalysts like Nvidia’s August print and each hyperscaler’s next capex update; and pre-commit to an exit trigger — the first hyperscaler cutting capex guidance matters more than any single drawdown.

Volume is intact. Valuation and funding are what will move these stocks from here.

---

This is a framework for thinking about the sector, not investment advice, and I’m not a licensed financial advisor.

For more, browse our full equity research reports, check the earnings calendar for what's coming up, or head back to the blog for the latest updates.