Amazon’s ‘Coiled Spring’ Moment: Can AWS’s AI Supercharge Ignite a $178B Breakout?

Amazon reports earnings tomorrow after the close, and this one matters. The stock has badly trailed the rest of Big Tech this year, AWS is under pressure to prove it can reaccelerate in AI, and management just announced 14,000 corporate job cuts to “move faster.” Investors are treating this print as a credibility check : can Amazon show that all the spending on AI infrastructure, data centers, retail automation, and advertising is actually translating into growth, margin, and cash — or is it just very expensive noise? Expectations are high but not euphoric: Wall Street is looking for revenue of about $178 billion (up ~12% year over year) and earnings per share of roughly $1.57, up ~10% from a year ago.

Let’s lay out the setup. After Q2, Amazon guided Q3 net sales to a range of $174 billion to $179.5 billion, implying 10%–13% year-over-year growth and a modest FX tailwind of ~130 bps. Management also guided Q3 operating income to $15.5–$20.5 billion. Street consensus is sitting near the high end on revenue and near the low end on operating income, which tells you the debate: top line is not the issue, margins are. AWS, ads, and cost discipline in retail need to carry the story enough to offset wage, logistics, depreciation, and tariff risk.

AWS is the heart of the call. Consensus has AWS net sales at about $32.3 billion, up ~18% year over year, which would be a slight acceleration from the 17.5% growth last quarter. That sounds fine… until you remember that Azure and Google Cloud are both growing north of 30% in AI-heavy workloads. The market no longer rewards AWS for “growing nicely.” It wants proof that AWS can bend back toward 20% growth and defend strategic relevance in AI. Analysts are very explicit about the trigger: if AWS can show a line of sight — even verbally — to 20%+ growth by 2026, sentiment turns. If not, Amazon keeps its “AI laggard” label.

To that point, Amazon has been rolling out narrative support ahead of the print. “Project Rainier,” its next-generation data center buildout, is now described as fully operational, with Anthropic using it to train and deploy Claude on what Amazon says will be more than 1 million Trainium2 chips across AWS by year-end. The message is: we’re not just renting Nvidia GPUs, we’re standing up our own AI-optimized infrastructure at hyperscale, with an anchor tenant. That’s also why multiple analysts — Citi, Wells, Mizuho — are back in the “AWS is about to reaccelerate” camp. They’re modeling AWS growth in the high teens now, crossing 20% in 2026 as Rainier capacity comes online, Anthropic scales spend, and AI inference workloads move from pilot to production. The flip side: UBS and others still see only modest upside near term and warn that AWS is giving up share to Microsoft and Google in AI-heavy deals, while margin pressure from capex and custom silicon could linger. Expect a lot of questions on backlog, capacity constraints, pricing, and whether AWS is still losing big AI workloads to rivals.

Just as important, AWS margins fell from a record 39.5% in Q1 to 32.9% in Q2, thanks to stock-based comp and heavier depreciation from all this new AI gear. Investors will want to know if that was a one-quarter reset or the new normal. Management has made the case that margins dip while it builds capacity, then recover as the racks fill. Tomorrow is where they have to defend that math.

Advertising is the quiet engine that’s propping up the rest of the model. Ad revenue grew 22% year over year in Q2 to about $15.7 billion, outpacing overall company growth and carrying extremely high incremental margins. Street is looking for roughly $17+ billion in ad revenue this quarter, which implies ~20% growth. The ad business sits at the intersection of retail data, Prime Video, connected TV, and off-Amazon partnerships (Roku, Disney, etc.). Analysts increasingly view ads as a structural margin lever: every ad dollar drops a lot cleaner than a dollar of first-party retail. Going forward, Amazon’s pitch is that ads are still under-monetized — Prime Video is just getting going on ads, and shoppable video plus grocery partnerships (for example, Winn-Dixie inventory inside Amazon with loyalty integration and local delivery economics) are designed to extend ad surfaces into everyday spend. That all feeds the “retail is getting leaner, higher-margin, and more defensible” argument bulls will push if AWS doesn’t blow the doors off.

Retail/commerce itself is expected to show solid but not spectacular growth. Analysts forecast Online Stores revenue of about $66.5 billion (+8% y/y), Third-Party Seller Services around $41.9 billion (+~11% y/y), and Subscription Services (Prime, etc.) at roughly $12.5 billion (+~11% y/y). Physical Stores — which includes grocery formats — is expected at about $5.6 billion (+6% y/y). The theme here is consistency: consumers are still spending, order frequency is holding up, and Amazon is taking wallet share without obvious tariff-driven price shock. Banks tracking card data say North America marketplace revenue is running ahead of consensus and there’s been “no notable increases in ASPs from tariffs so far,” which is exactly what investors wanted to hear heading into holiday guidance.

But this is not a cost-blind growth story anymore. Amazon just announced ~14,000 corporate job cuts — about 4–5% of white-collar headcount — framed as “fewer layers, faster decision-making,” not as distress. Citi pegs potential annual savings at $3.5–$4.2 billion on top of ongoing logistics automation and fulfillment network redesign. Management will almost certainly be asked whether these cuts mean retail is slowing or whether this is Amazon rediscovering vintage “Day 1” discipline and using AI/automation (robotics in fulfillment, route optimization like DeepFleet, etc.) to structurally lift margins. That goes directly to operating income guidance for Q4 and into 2026.

We also need to talk about Q2, because Q2 sets the bar. Last quarter, Amazon posted $167.7 billion in revenue, +12% year over year, beating expectations; EPS beat as well, with operating income up more than 30% year over year. AWS grew 17.5% to $30.9 billion in revenue, ads grew ~22%, and both North America and International delivered margin expansion as faster delivery, regionalized logistics, and automation kicked in. Prime Day was flagged as a record on basically every metric that matters (items sold, new Prime sign-ups). Management bragged about reducing average delivery distance ~12% and cutting handling touches per unit ~15%, which translates to real unit cost leverage at scale. In plain English: retail isn’t just growing, it’s getting cheaper to run.

Guidance coming out of Q2 for this Q3 call was deliberately cautious. Management said Q3 operating income could be as low as $15.5 billion and openly admitted tariffs are an uncertainty they “just don’t know” how to model. ( IG) They also said AWS demand is ahead of capacity — which sounds bullish, but also implies elevated capex and lower near-term margin because Amazon has to build data centers, power infrastructure, and custom silicon before it can fully monetize AI workloads. That’s one of the reasons bears argue Amazon is spending heavily just to “catch up” in AI, not to lead it.

So what are the primary things to watch tomorrow?

AWS growth rate and tone. Can management credibly talk about acceleration toward 20%+ growth, not just “we did 18% this quarter”? Do they lean on Project Rainier, Anthropic, and backlog to make that case?

AWS margins. Was Q2’s drop to ~33% operating margin a trough or the start of structurally lower profitability as Amazon funds AI capacity?

Ad momentum. Do they confirm ~20%+ ad growth and frame Prime Video ads / retail media / off-Amazon DSP deals as multi-year, high-margin expansion?

Holiday/Q4 guide. Street is braced for a conservative outlook on Q4 revenue and profit, given tariff noise and regulatory overhang (including the recent FTC settlement). Amazon historically lowballs, but sentiment in this stock probably can’t absorb another “cautious” quarter unless AWS truly shines.

Cost discipline. Does management explicitly tie the 14,000 job cuts and robotics/automation rollout to multi-billion-dollar operating income lift in 2026?

Bottom line: The bar has quietly moved. Beating revenue and EPS tomorrow is not enough. Amazon has to sell a story: AWS is reaccelerating with AI capacity (and will be paid for it), ads are scaling into a second profit engine, retail is more automated and less tariff-sensitive, and cost discipline is real. If they can check most of those boxes, the stock finally has room to act like the “ coiled spring ” the bulls keep promising.