Amazon reports second-quarter results on July 30, one week after Alphabet delivered the clearest warning yet for the AI trade. This Amazon Earnings Preview begins with a simple tension: Google’s parent posted exceptional cloud growth and better-than-expected revenue, but the stock still weakened after management raised its 2026 capital-expenditure forecast to $195 billion–$205 billion and reported negative quarterly free cash flow of $5.9 billion. The market’s message was difficult to miss: strong AI demand no longer excuses unlimited spending.
Amazon now faces the same test with greater financial tension. Wall Street expects roughly $197 billion to $199 billion of quarterly revenue, improving retail margins and another acceleration at AWS. The obvious debate is whether AWS can grow above 30% as Anthropic, OpenAI and enterprise customers consume more AI infrastructure.
But that may not decide the reaction.
The real earnings question is whether Amazon can show that operating profit is beginning to catch up with its enormous infrastructure bill. AWS growth matters this quarter. The relationship between capital expenditure, operating income and free cash flow will matter over the next two quarters.
What Wall Street Is Modeling In The Amazon Earnings Preview
Amazon’s own guidance calls for second-quarter sales of $194 billion to $199 billion, representing growth of 16% to 19%. Management also expects operating income between $20 billion and $24 billion, compared with $19.2 billion a year earlier.
Expectations have moved toward the upper end of those ranges. Bank of America projects revenue of $198.8 billion and operating profit of $24.1 billion, both modestly above broader consensus. That leaves Amazon with a familiar earnings challenge: merely reaching management’s ceiling may not feel like a meaningful surprise.
AWS carries even more demanding expectations. After growing 28% in the first quarter—its fastest rate in 15 quarters—some analysts now expect second-quarter growth near 33%, followed by further acceleration during the second half. The narrative is supported by rising Bedrock consumption, new OpenAI availability and substantial commitments from Anthropic.
Margins are expected to remain unusually strong as well. BofA models an AWS operating margin near 34%, helped by high utilization, pricing and the increasing use of Amazon’s custom chips. Retail is supposed to provide additional stability, with North American sales growth supported by Prime Day and continued fulfillment-network efficiencies.
That is a constructive setup, but also a crowded one. Investors already understand that AWS demand is strong, Trainium capacity is heavily committed and Amazon’s retail network is becoming more productive.
The consensus narrative assumes that faster AI growth will eventually solve the spending problem. Earnings must provide evidence that “eventually” is getting closer.
The Metric That Actually Matters
The most important number will not be AWS revenue growth in isolation. It will be the incremental operating return on Amazon’s AI capital expenditure.
Amazon spent $43.2 billion in cash capital expenditures during the first quarter, primarily on AWS and generative AI. Management has explained that the company must fund land, power, data centers, networking equipment and chips six to 24 months before much of that infrastructure begins generating revenue.
That timing mismatch is central to the Amazon Earnings Preview. Amazon can post accelerating AWS revenue and record operating income while free cash flow remains deeply pressured because cash leaves the business well before customers fully consume the new capacity.
The problem is now too large to treat as a temporary accounting detail. Amazon’s trailing-12-month free cash flow fell from $25.9 billion to approximately $1.2 billion in the first quarter as purchases of property and equipment surged. The company has argued that these assets have long useful lives and should generate attractive cumulative returns once utilization scales. That may prove correct, but the market increasingly wants proof rather than duration-based assurances.
Management has several ways to provide it. AWS operating income could grow faster than revenue. Trainium could reduce the cost of serving inference workloads. Customer commitments could shorten the time required to fill new capacity. Retail and advertising profits could also absorb more of the investment burden.
The mechanism is straightforward: when operating income rises faster than the infrastructure base, investors can model a future free-cash-flow rebound with greater confidence. When capital expenditure rises faster than operating income, the terminal value of that spending becomes less certain and the valuation multiple becomes harder to defend.
That is why Amazon’s potential 2026 capital-expenditure update could matter more than a modest revenue beat. BofA believes the company could lift its outlook toward $210 billion because of stronger demand and higher memory costs. After Alphabet’s reaction, a larger spending plan without a parallel improvement in monetization would be difficult for investors to dismiss.
Amazon does not need to reduce AI spending; it needs to show that each new dollar is producing visible operating leverage.
The Upside Case: AWS Turns Spending Into Earnings
The clean upside scenario begins with AWS growth above the market’s elevated expectations. Growth around 33% would confirm that the first-quarter acceleration was not simply a comparison benefit and that AI workloads are moving from experimental spending into sustained production demand.
The quality of that growth would matter more than the headline. Investors would want to hear that demand is broadening beyond a small group of frontier-model developers and that traditional enterprises are increasing both AI consumption and core cloud usage. Amazon has already said that AI adoption creates demand for storage, databases, analytics, CPUs and other AWS services, giving the company an opportunity to monetize the same customer across a wider stack.
Trainium could then become the key margin bridge. Management says Trainium2 offers roughly 30% better price performance than comparable GPUs, while Trainium3 improves further on Trainium2. Amazon also expects custom silicon to save tens of billions of dollars in annual capital expenditure at scale and provide several hundred basis points of operating-margin advantage for inference workloads.
Those claims will carry more weight if AWS margins remain resilient while AI revenue accelerates. A margin near or above 34%, despite higher stock-based compensation and substantial infrastructure additions, would suggest that utilization and custom-chip economics are offsetting the cost of expansion.
Advertising could strengthen the upside narrative. The business generated $17.2 billion of first-quarter revenue, up 22%, and its high incremental margins can help subsidize investments elsewhere. Retail efficiency would provide another support if fulfillment and shipping costs continue growing more slowly than units.
In this Amazon Earnings Preview upside scenario, investors may look through weak current free cash flow because the route toward recovery becomes more credible. The market would begin to treat elevated capital spending as contracted capacity for visible demand rather than an open-ended wager on future AI adoption.
The bullish surprise is not simply faster AWS growth—it is faster AWS growth without proportional margin deterioration.
The Downside Case: The Capex Hike Overwhelms The Beat
The more dangerous scenario is not an outright AWS miss. It is a superficially strong quarter that fails the same test Alphabet just failed.
Amazon could exceed revenue estimates, deliver AWS growth above 30% and still disappoint if management raises capital-spending expectations sharply while offering limited evidence of improving returns. Alphabet demonstrated that the market is willing to punish a hyperscaler even when cloud demand is exceptional if free cash flow deteriorates faster than expected.
Memory and storage costs add another layer of risk. Amazon acknowledged last quarter that component prices had risen sharply because supply was insufficient for industry demand. The company has secured meaningful supply through strategic partners, but higher input costs can still inflate the cash required to build each unit of computing capacity.
AWS margins could therefore become the first visible pressure point. If growth accelerates but operating margins fall more than expected, investors may conclude that new AI revenue is less profitable than the mature cloud workloads it is replacing at the margin. Heavy depreciation, power expense, stock compensation and third-party chip purchases could all restrict operating leverage.
Customer concentration is another potential concern. Amazon reported a $364 billion AWS backlog at the end of the first quarter, excluding a subsequently announced Anthropic agreement worth more than $100 billion. It also disclosed more than $225 billion of Trainium revenue commitments.
A large backlog is not the same as immediate cash generation. Revenue may be recognized over many years, while Amazon incurs much of the infrastructure cost upfront. Commitments from well-funded AI laboratories are valuable, but investors may still discount them if capacity requirements keep moving higher or contract margins remain unclear.
Amazon Leo could further complicate the picture. The satellite business is expected to create approximately $1 billion of year-over-year cost pressure in North America during the second quarter, with commercial service scheduled to begin in the third quarter. Leo may become a valuable long-duration asset, but it represents another capital-intensive project competing for cash while AWS is already consuming record investment.
The downside case in this Amazon Earnings Preview is therefore less about weak demand and more about deteriorating capital efficiency.
The downside break occurs when investors decide Amazon is adding capital commitments faster than it is adding economic returns.
What Matters After The Print
The July report will provide a snapshot, but Amazon’s AI economics will be judged across the next two earnings calls.
The first signal is the progression of AWS growth. A second-quarter acceleration followed by stronger third-quarter guidance would indicate that installed capacity is being monetized quickly. A strong quarter followed by cautious growth commentary would raise the possibility that large AI contracts are producing uneven rather than durable acceleration.
The second signal is AWS margin stability. Custom silicon should become increasingly important as Trainium3 deployment expands and Bedrock processes more inference internally. If AWS margins remain in the mid-30% range during that transition, Amazon’s argument that vertical integration creates an economic advantage becomes much stronger.
The third signal is the relationship between operating cash flow and capital expenditure. Free cash flow may remain weak for several quarters, but the direction of the gap matters. Investors will look for evidence that operating cash generation is beginning to rise fast enough to absorb the infrastructure program rather than allowing spending to compound indefinitely.
The competitive backdrop will also sharpen. Alphabet’s Google Cloud has accelerated dramatically, Microsoft continues building around Azure and OpenAI, and Nvidia remains the default platform for many high-end workloads. Amazon’s advantage depends on offering customers model choice, infrastructure reliability and lower-cost silicon without limiting access to Nvidia systems.
Anthropic adds both strategic value and financial noise. Amazon benefits from Anthropic’s AWS consumption and from changes in the valuation of its investment, but mark-to-market gains can distort reported net income. Investors should therefore focus on segment operating income and cash generation rather than treating GAAP earnings as a clean measure of underlying performance.
Advertising and retail deserve continued attention as well. Both businesses can cushion the AI investment cycle, but they cannot permanently compensate for weak infrastructure returns. Their role is to preserve consolidated earnings power while AWS capacity matures—not to make the economics of that capacity irrelevant.
For the Amazon Earnings Preview to age well over the next year, contractual demand must become measurable cash generation.
Over the next 6–12 months, Amazon must convert contractual AI demand into utilization, utilization into margins and margins into cash.
The Earnings Interpretation
Amazon enters the quarter with most of the favorable operating narrative already established. AWS is accelerating, AI demand is strong, Trainium is heavily subscribed, advertising is growing and the retail network is more efficient.
The unresolved issue is whether those strengths are improving the company’s cash-return profile quickly enough to justify an infrastructure program approaching $200 billion—or potentially more. A revenue beat would confirm demand. Strong AWS margins would validate execution. A credible path toward free-cash-flow recovery would address the actual debate.
Valuation makes that distinction important. Amazon’s consensus earnings multiple reflects expectations that current spending will create a larger, more profitable cloud and AI platform rather than a permanently more capital-intensive business. Multiple expansion requires increasing confidence in future returns; multiple compression becomes more likely when the required investment rises faster than the evidence supporting those returns.
Alphabet has already shown that spectacular cloud growth cannot fully insulate a hyperscaler from capital-efficiency concerns. Amazon’s report will reveal whether AWS has different economics—or whether it is simply the next company facing the same reckoning.
The central conclusion of this Amazon Earnings Preview is that the headline beat matters less than the evidence supporting future returns.
After July 30, the number to monitor will not be how much Amazon spends, but how quickly that spending begins paying Amazon back.
Disclaimer: We do not hold any positions in the above stock(s). Read our full disclaimer here.
