Amazon isn't the cheapest of the Magnificent 7. But it has something that's worth more than a low P/E
Over the past few months, one sentence keeps popping up on social media: Amazon is the cheapest stock of the Magnificent 7. It reads well and shares even better. But the numbers don't add up. Based on estimated P/E, Amazon was only third at the beginning of October. Still, I think the people who claim it aren't entirely off. They're just right for a different reason than they think.
What the numbers show
I'll start with forward P/E. That's the stock price divided by the earnings analysts expect over the next twelve months. As of today, Meta is the cheapest at 22.1 times. Alphabet follows at 22.8, and only then Amazon at 23.6. Nvidia is at 24.9 and Microsoft at 25.1. Apple is over 35 and Tesla at roughly 159 plays a completely different game. 😄
It's worth noting how small the differences are among the top five. Between Meta and Microsoft, there are three P/E points. The title of "cheapest" is therefore decided by a few tenths and can change after any earnings release.
When I add growth to the price, the picture flips even more. The PEG ratio divides P/E by expected earnings growth. By that measure, Nvidia looks cheapest at 0.48. Meta is around 1, Alphabet 1.26, and Amazon 1.48. According to this metric, Amazon is fourth.
So where does that narrative come from? Mainly from historical P/E around 20. But that's distorted. In this year's results, Amazon booked roughly 53 billion dollars from revaluing its stake in Anthropic. That's an accounting profit, not money the e-commerce or cloud business actually brought in. When I subtract it, Amazon stops looking cheap at first glance.
Two companies in one coat
Amazon has an operating margin of 13.7%. Alphabet has 34% and Microsoft over 45%. At first glance, Amazon looks like the weakest business of the group. But retail—warehouses, logistics, and shipping—drags down its overall margin.
AWS grew 36.7% in the second quarter, its fastest rate in 18 quarters. The cloud operating margin was 39.4%. By my calculation, AWS makes up about a fifth of revenue but over 60% of the company's operating profit. The cloud's contractual backlog, meaning orders not yet reflected in revenue, is 496 billion dollars.
The second driver is advertising. It brought in 19.8 billion dollars for the quarter and grew 26%. It's a high-margin business built on data about what people actually buy.
Total revenue grew 19.6%, but operating profit grew 43%. Profit is thus growing more than twice as fast as revenue. I'm paying 23.6 times earnings for a company whose mix is slowly shifting from low-margin retail to high-margin cloud and advertising. That's why I think Amazon doesn't have the lowest price tag, but it offers the best ratio of price to what I get for it.
Where's the catch
I won't pretend there isn't another side. Capital expenditures (capex) were 54.2 billion dollars in the second quarter, up 68% from a year ago. For the full year, roughly 200 billion is expected, and according to the latest estimates even more. Free cash flow over the last twelve months is therefore negative, around minus 7.6 billion dollars.
Based on the money the company actually has left, Amazon isn't cheap at all. I'm betting that those hundreds of billions in data centers will pay off. The 496 billion backlog is the strongest argument so far that they will.
I'm not buying the price tag, but the direction
I don't look at Amazon through whether its P/E is one point lower than Meta's. I look at where its profit is shifting. And it's moving one way: from boxes and deliveries to cloud and advertising. Every quarter that AWS and advertising grow faster than e-commerce, Amazon becomes a more profitable company without having to raise prices.
Meta is cheaper by P/E, but almost all of its profit rests on advertising. Alphabet is cheap, but the market still hasn't forgiven its fear that AI will eat into search. Amazon has three drivers at once: cloud, advertising, and retail, which itself acts as a massive data source for the other two. For that diversification, a few tenths of extra P/E don't bother me.
I treat the Anthropic stake as a bonus. I don't count it in valuation. If it works out, it's a cherry on top. If not, the thesis stands without it.
What do I watch as a warning sign? The ratio between AWS growth and capex growth. As long as cloud accelerates and the backlog grows, those billions in data centers make sense. But if capex continued to rise and AWS started to slow, it would stop being a story about investment and start being a story about spending. That's where I would reconsider my view.
