Big Tech Q2 2026 Earnings Preview: The AI Capex Test

The most consequential stretch of the earnings calendar is about to begin, and this quarter the market is grading on a stricter curve. Between July 22 and July 30, 2026, the five companies at the center of the AI trade — Alphabet, Microsoft, Meta, Amazon, and Apple — will open their second-quarter books. The numbers investors care about most are not revenue or earnings per share. They are the capital-expenditure lines, and whether the businesses those dollars are building can yet show a return. Collectively, the hyperscalers have guided toward roughly $700 billion in 2026 capital spending, and Wall Street has made clear it now wants to see the receipts.

The calendar

The reporting schedule front-loads the season’s biggest questions. Alphabet kicks things off on Tuesday, July 22, making it the first of the group to face the capex interrogation. Microsoft and Meta both report after the close on Wednesday, July 29. Apple and Amazon close out the week on Thursday, July 30. In the span of nine days, companies worth well over $15 trillion combined will lay out how much they are spending, how fast that spending is growing, and — the part that matters most — what it is earning.

The number that defines the season

The headline figure is the capital budget, and it is staggering. Guidance across the four hyperscalers points to combined 2026 capex near $700 billion, roughly double what the group spent in 2025. Broken down, the individual commitments are without precedent:

• Amazon leads the pack at approximately $200 billion, aimed largely at data-center expansion and custom silicon.
• Microsoft is tracking toward roughly $190 billion for the calendar year.
• Alphabet has guided to a range in the neighborhood of $175 billion to $185 billion.
• Meta has repeatedly revised its guidance upward, most recently to a band around $115 billion to $135 billion, citing higher component and data-center costs.

Add Oracle and the rest of the buildout, and some analysts now put the top-five figure closer to $750 billion, up sharply from estimates near $620 billion at the start of the year. The overwhelming majority of that money flows to AI data centers, Nvidia GPUs, custom accelerators, and the electricity to run them. This is the hyperscaler capital-spending boom in its most extreme form, and every one of these companies is racing at once on the logic that being short on compute is more dangerous than overspending.

Why the mood has changed

For most of the AI cycle, capital spending was a feature, not a bug. Announcing bigger budgets signaled conviction, and the stocks rose accordingly. That reflex has broken down. The era of rewarding a company simply for invoking artificial intelligence is over; investors are now demanding concrete evidence that the spending is converting into revenue, margin, and cash flow.

The reason is arithmetic. Capital spending of this magnitude does not vanish — it lands on the balance sheet as assets that must be depreciated, which drags on reported earnings for years, and it consumes cash that would otherwise fund buybacks and dividends. When spending doubles while revenue grows at a more modest clip, the gap between the two widens, and the market has started to notice. The economics of the AI data-center buildout only work if utilization and pricing hold up long enough to earn back the outlay, and a single quarter of soft cloud growth would reopen every question the bulls thought they had answered.

That scrutiny is exactly what drove the recent repricing of the AI trade. It is the same anxiety that pushed Apple past Nvidia to reclaim the most-valuable-company crown on July 17 — a market rewarding the company with the lightest capital budget and the heaviest free cash flow, and punishing the names most exposed to the spending cycle.

What to watch at each company

The five reports test different pieces of the same thesis.

Alphabet (July 22). As the opener, Google sets the tone. The cloud segment’s growth rate and margin will be read as a proxy for whether AI demand is translating into paid workloads. Any commentary on capex trajectory — raise, hold, or moderate — will move the entire group, given Alphabet reports first.

Microsoft (July 29). The purest read on enterprise AI monetization. Azure’s growth rate, the contribution attributed to AI services, and the pace of capital spending are the three lines that matter. Microsoft has flagged that a meaningful slice of its capex reflects component pricing rather than pure capacity, a nuance investors will want quantified.

Meta (July 29). Meta is spending like a hyperscaler while monetizing through advertising, which makes its capex the hardest to underwrite. The company has raised guidance repeatedly, and investors will want to see AI-driven engagement and ad performance that justifies a budget approaching that of the cloud providers. Its recent move to lease compute to Anthropic adds a new wrinkle: infrastructure as a revenue line, not just a cost.

Amazon (July 30). The largest spender of all. AWS growth is the swing factor for the entire cloud narrative, and Amazon’s custom-silicon strategy is central to its pitch that it can build cheaper than it buys. With roughly $200 billion earmarked for the year, the market wants evidence the capacity is being filled with paying demand.

Apple (July 30). The outlier by design. Apple is not a hyperscaler and does not carry a comparable capital budget, which is precisely why the market has re-rated it. The questions here are different: services growth, the pace of AI feature adoption across its device base, and any read on the China recovery. Apple closes the week as the counter-example to everyone else’s spending story.

The bigger picture

Not every AI-adjacent print this season has been reassuring. A guidance cut can still rattle a stock hard, as Netflix’s cautious outlook showed, and the upstream chip names sold off even on TSMC’s blowout results. The pattern rhymes: the bar has risen, and good-but-not-great is no longer good enough when valuations assume years of uninterrupted growth.

What it means

This is the quarter where the AI capital-spending thesis meets its first hard, collective test — five companies, nine days, one question.

The stakes. With roughly $700 billion committed for 2026, even a small shortfall in monetization changes the math on the whole buildout. The market is no longer paying for ambition; it is paying for proof. Reports that pair heavy spending with visible AI revenue and holding margins will be rewarded. Reports that show spending outrunning returns will not.

Who is most exposed. The heaviest spenders — Amazon, Microsoft, Alphabet, and Meta — carry the most risk into their prints, because their valuations already embed a payoff that has yet to fully arrive. The lighter-capex beneficiaries, Apple chief among them, enter with the market’s goodwill precisely because they are not making the same bet.

What to watch next. First, Alphabet on July 22 — as the opener, its capex commentary and cloud growth will frame every report that follows. Second, whether any of the four hyperscalers blinks and trims guidance, which would be read as the first crack in the buildout. Third, the cash-flow lines, not just earnings, because that is where record capex actually shows up. The spending is locked in. Over the next nine days, the market finds out what it is buying.