Record margins, AWS’s fastest growth in 5 years, stock down 6%
Over the past 2 days, all 3 big companies reported earnings: Microsoft $MSFT rose after its results, Alphabet $GOOG lost over 12%, and Amazon plunged as much as 10% right after the market close before recovering some of its losses. Yet it reported the most profitable quarter in its history. To understand why the market is reacting this way, we need to look at how Amazon $AMZN operates today and where its money comes from. It’s not just the traditional retailer we all know.
Amazon is 3 companies in 1
At first glance, Amazon is an online store. Financially, though, it’s something entirely different:
Retail (North America and International segments) brought in $156 billion in revenue, or about 77% of the total. But it accounted for only about 37% of operating income. That alone is something worth taking a closer look at.
AWS (cloud) brought in $46.0 billion, or “just” 23% of revenue, but generated 63% of the company’s total operating income. So today, Amazon is clearly making most of its money from the cloud. That’s why the market is paying so much attention to it.
Advertising (sponsored products on Amazon, ads on Prime Video and sports broadcasts) generated $19.6 billion. The company doesn’t report its profit separately, but it has the highest profit per dollar Amazon earns. It’s paid for by merchants who are already selling on Amazon, so costs are minimal. That’s important too. It’s what makes it so profitable.
Amazon’s stock doesn’t move based on how many people buy something on Amazon.com, but on AWS and advertising. So let’s look at those businesses, because that’s where things get interesting.
AWS
AWS grew 39%, its fastest pace in almost 5 years.
Its annualized revenue run rate reached $184 billion. For comparison, a year ago AWS cloud growth was “only” 20%, and the market feared it was losing the cloud AI race to Microsoft and Google.
Its growth rate doubled in 12 months. For a business that’s already the world’s largest cloud provider, every bit of additional growth is simply harder to achieve. By contrast, Google $GOOG, whose stock plunged so dramatically today (it was Google’s biggest stock drop since its 2004 IPO), is growing its cloud business much faster. Google Cloud grew 103% year over year. That’s insane, but it still wasn’t enough for the market.
The bigger a company gets, the harder it is to grow quickly. Amazon’s cloud added $13 billion in revenue year over year in a single quarter. That’s undeniably solid.
Andy Jassy (Amazon’s CEO) also gave a fairly bold outlook, at least in my view.
He expects AWS to grow by more than 40% in both 2027 and 2028. But he wouldn’t dare give a number like that without signed contracts. So the backlog is just as important here as it is for Microsoft or Google. Management said back in the summer that much of its 2027 capacity was already reserved. That alone is positive.
AWS operating income was $17.8 billion at a 38.7% margin, compared with 39.4% in the previous quarter. I focused on this, even if you might say the decline is small.
A data center doesn’t show up in the results on the day Amazon pays for it, but gradually through depreciation spread over several long years. So today’s huge investments will keep putting pressure on AWS margins long after the company spends the money. And of course, that’s not unique to Amazon. If AWS can maintain a margin of around 38 to 39% even as depreciation rises, that will be the best proof that the investments are paying off.
Trainium: a quiet weapon against Nvidia
Amazon’s own chips provide a big part of the answer to how it can maintain margins. The market can appreciate that, too.
Most AI companies pay Nvidia $NVDA for computing power, and Nvidia maintains margins of around 70%. Amazon designs its own AI chips, Trainium, and Graviton processors.
Computing workloads running on Amazon’s own chips instead of Nvidia’s are significantly cheaper for its cloud business. The custom chip business was already running at an annualized rate of over $25 billion in the summer, and both Anthropic and OpenAI have made multiyear commitments to Trainium with Amazon. We’re talking about their computing capacity in gigawatts. Amazon is even considering selling Trainium outside its own cloud, much like Google has started doing with its TPUs (tensor processing units).
How AWS stacks up against the competition
This week, all 3 companies with cloud services reported earnings:
AWS: $46.0 billion, up 39%
Microsoft Azure: $33.2 billion, up 48%
Google Cloud: $30.7 billion, up 103%
AWS is still the biggest, but it’s growing the slowest. That was expected, so it’s no surprise. But if we look at absolute revenue growth, Google came out on top. Google Cloud generated $15.5 billion more year over year, more than AWS. The market leader is accelerating, but the competition is accelerating even faster. A quick aside: now imagine if Google Cloud were growing “only” 50%… those stocks probably wouldn’t find a bottom 😅
Record margins, but watch out for comparisons
The company’s total operating income reached $28.1 billion and its operating margin was 13.9%, the highest in Amazon’s history. The company also beat the top end of its own guidance, which was $26.5 billion.
The 61% increase in profit, as stated in Amazon’s data, is a little misleading, though. Last year’s third quarter was weighed down by $2.5 billion in a settlement with the U.S. regulator FTC and $1.8 billion in severance costs. Excluding those, operating income grew by roughly 30%. That’s very good too, but today it’s simply not enough for the market…
Operating income in North America rose to $8.7 billion from $4.8 billion, and the margin climbed from roughly 4.5% to 7.7%. Some of that is due to last year’s one-time costs, but the trend is clear. In recent years, Amazon has restructured its logistics into a regional network of warehouses, so goods travel shorter distances, and it has deployed hundreds of thousands of robots in its warehouses. The e-commerce business, which for a long time just burned money to grow, is gradually turning into a reasonably profitable business. Another positive development.
Total revenue grew 12% to $201.9 billion. It’s worth knowing that Amazon moved Prime Day from July to June this year. That big sale event therefore fell in the second quarter, whereas last year it was in the third. Without that shift, revenue would have grown about 4 percentage points faster.
3 reasons $AMZN stock fell
1. Advertising growth halved
Advertising revenue reached $19.6 billion, while the market expected $20.8 billion.
Growth slowed to 12% from 26% in the previous quarter. Amazon blames the timing shift of Prime Day and promises a return to growth of around 15 to 19% in Q4. But the market is also nervous for a longer-term reason. If people shop through AI assistants that choose products for them, they’ll bypass sponsored results that merchants pay Amazon for. That’s why Amazon blocked Meta’s $META shopping AI agent from its website in September. Muse has ended up at odds with Amazon. Well, I’m curious to see what happens with Meta, too.
2. Earnings per share missed expectations
EPS was $1.89, versus expectations of $1.96.
But it’s not that simple; let me explain: Amazon owns a large stake in Anthropic, and whenever its valuation changes, Amazon has to revalue its stake and record the difference in earnings.
That kind of revaluation brought Amazon $9.5 billion last year in the third quarter. In this year’s second quarter, it brought in a whopping $53.4 billion. EPS jumped to $5.75 as a result. That bonus was missing this year in the third quarter, so EPS fell year over year, even though the operating business earned significantly more. With Amazon, as with Alphabet and SpaceX $SPCX, it’s therefore better to track operating income.
3. Cash flow is negative (just like Google’s)
This is the main problem. Over the past 12 months, Amazon generated a record $168.3 billion in operating cash flow. Free cash flow, or what’s left after paying for investments, was -$42.4 billion, though. A year ago it was +$14.8 billion. From that, we can calculate that Amazon invested roughly $211 billion over the past year, $43 billion more than it earned.
For the full year 2026, it plans to invest around $220 billion. That’s BTW more than AWS brings in over a year. Amazon is financing the difference with debt. Long-term debt doubled in the first half of the year, from $65.6 billion to $128.9 billion.
Outlook: Q4 looks strong
For the fourth quarter—the Black Friday and Christmas season—Amazon expects revenue of around $243.5 billion and operating income of $28.5 to $33.5 billion, compared with $25.0 billion last year.
It hasn’t changed its investment outlook for 2026. That’s positive. But the main question is what happens in 2027. The competition, especially Google, has already signaled a significant increase. Current estimates suggest hyperscalers could spend over $1 trillion combined next year. That’s insane.
I have the stock in my portfolio. I’m not buying or selling, just holding. I bought in February after earnings.
So, what do you think? Are you buying the dip, or are you looking at your portfolio with disdain right now?
