Amazon Q2 2026 Earnings Preview: AWS, Capex, Chips

The last of the megacaps steps up to the plate on Thursday. Amazon is scheduled to report second-quarter 2026 results after the market close on July 30, alongside Apple, closing out a compressed earnings week that has already reshaped how investors are pricing the artificial intelligence buildout. After Alphabet, Microsoft, and Meta each delivered strong numbers and got a wary reception, the bar for Amazon is simple and unforgiving: prove the cloud is reaccelerating fast enough to justify a capital-spending plan that is now measured in the hundreds of billions.

The stock arrives at the print under pressure. Shares in Amazon (AMZN) have fallen roughly 11% over the past three months, weighed down by the same anxiety hanging over the whole sector — that the industry is committing to more data-center capacity than it can profitably fill. Analysts, for their part, remain broadly constructive, but the reaction to this quarter will hinge less on whether Amazon beats and more on the two lines the market now grades hardest: AWS growth and capital expenditures.

The headline numbers Wall Street expects

Consensus estimates cluster tightly. Revenue is seen near $196.9 billion, up about 18% year over year, with adjusted earnings per share around $1.82, an increase of roughly 8%. Amazon’s own guidance for the quarter called for operating income of $20 billion to $24 billion, compared with $19.2 billion a year earlier — a range wide enough to accommodate both the profit engine of the retail and advertising business and the margin drag from a heavier investment cycle.

Those are healthy figures. But headline beats have not been rewarded this season. The season’s opening act, Alphabet, beat on revenue and showed Google Cloud growing 82% — and the stock fell anyway once management lifted its capex guidance. The lesson carried straight through Microsoft and Meta: this quarter, the market is grading the spending, not the top line.

AWS is the whole story

For Amazon, the number that matters most is AWS. The cloud unit is expected to generate roughly $40.5 billion to $40.6 billion in the quarter, about 31.6% growth year over year. That would mark a meaningful acceleration from the 28% it posted in the first quarter — itself the division’s fastest growth in fifteen quarters. AWS is the profit center that funds Amazon’s ambitions, and its growth rate is the single cleanest read on whether enterprise AI demand is translating into cloud consumption or merely into press releases.

The bull case is that AWS is finally shaking off the optimization slowdown that dogged it in prior years and is now capacity-constrained rather than demand-constrained — the same argument Microsoft made about Azure. If AWS clears 32%, it validates the idea that the spending has a customer on the other end. If it stalls near 28% again, the capex line becomes much harder to defend.

The capex line under the microscope

Amazon has been unusually direct about how much it intends to spend. During its fourth-quarter 2025 call, the company committed to raising capital expenditures by about 52% year over year, to roughly $32.9 billion in a quarter, and it held that line through the first quarter. Now the Street expects the guidance to move higher again. Wedbush projects Amazon will lift full-year 2026 capex above $200 billion and pencils in as much as $266 billion for 2027, with AWS-specific spending potentially reaching $165 billion in 2026 — up more than 70% on the prior year.

Those are staggering sums, and they land on the income statement long before the revenue they underwrite. Depreciation on tens of billions of dollars of servers, custom silicon, buildings, and networking flows straight through to operating margins. Consensus sees full-year AWS operating margin slipping about 50 basis points, but a steeper decline of 150 basis points is plausible once higher memory prices, a shifting revenue mix, and the AI investment ramp are accounted for. It is the same tension at the heart of the entire hyperscaler capex boom: the spending is real, the demand behind it appears real, and the payback is a matter of faith until it shows up in free cash flow.

That last metric will get outsized attention. Meta’s quarter was punished in part because record capex compressed free cash flow to a sliver of the prior year’s. Amazon funds much of its buildout from operating cash, but the market will be watching how much cash is left after the machines are paid for — and the company’s recent $25 billion AI bond sale is a reminder that even Amazon is reaching into debt markets to keep pace.

The chip business hiding inside AWS

The most underappreciated line in Amazon’s story is silicon. The company’s custom-designed chips — the Trainium training accelerators and Inferentia inference parts — now run at roughly $20 billion in annualized revenue and are growing at triple-digit rates, expanding nearly 40% quarter over quarter. On a standalone basis, some analysts argue the chip franchise could reach $50 billion selling this year’s output.

That matters for two reasons. First, every workload Amazon can move onto its own accelerators is a workload it does not have to buy Nvidia GPUs for, which improves the unit economics of AWS’s AI services and softens the margin hit from all that capex. Second, it gives Amazon a differentiated pitch to cost-conscious customers: cheaper inference on chips Amazon controls end to end. Any color management offers on Trainium adoption, supply, and the next generation will be read as a proxy for how defensible AWS’s AI margins really are.

Advertising rounds out the bull case. The ads business has been a standout, running around $17 billion a quarter and growing better than 20% year over year — high-margin revenue that helps offset the investment cycle and is increasingly central to how Amazon monetizes its retail traffic.

What to watch on the call

• AWS growth rate. Above 32% is a win; a print near 28% reopens the capex debate.
• 2026 and 2027 capex guidance. The market wants to see spending tied explicitly to backlog and demand, not aspiration.
• AWS operating margin. How much are AI investment and memory costs eroding the cloud’s profitability?
• Trainium and Inferentia commentary. Adoption, supply, and roadmap signal how much AWS can insulate its AI margins.
• Free cash flow and the debt picture. After Meta’s collapse in free cash flow, this line is now a headline metric.
• Fourth-quarter guidance. Amazon’s holiday-quarter outlook lands into a jittery correction-watch tape.

What it means

Amazon closes the most consequential earnings stretch of the year, and it does so as the swing vote on a question the market has been asking all week: is the AI buildout an investment or an overhang? The three megacaps that reported before it all told the same story — demand is strong, spending is going higher, and the stock fell anyway. Amazon has a chance to break that pattern, because unlike its peers it can point to two things they cannot match in combination: an AWS franchise that appears to be genuinely reaccelerating, and a custom-silicon business that lowers the cost of serving AI at scale.

If AWS clears the low-30s and management frames its capex as demand-backed and chip-optimized, Amazon can make the most credible case in the group that the spending pays for itself — and its beaten-down stock has room to rally. If AWS disappoints or the capex number climbs without a matching demand signal, Amazon will simply be the fourth megacap in a week to prove that, in the summer of 2026, investors have stopped taking “the demand is real” on faith.

Either way, the read-through extends far beyond Amazon. As the largest single buyer of its own AI infrastructure and a top customer for the whole semiconductor supply chain, what Amazon guides to on Thursday night sets the tone for chipmakers, memory suppliers, and every company betting that the cloud will keep absorbing capacity as fast as it can be built. The retail giant reports after the bell; the market’s verdict on the AI trade arrives with it.