Wall Street has a fantasy. It goes like this: Amazon spins off AWS, and suddenly the cloud giant trades at a pure-play tech multiple. Shareholders get a windfall. Regulators get their pound of flesh. Everyone goes home happy.
It’s a beautiful story. It’s also completely wrong.
Here’s what nobody on the sell-side is telling you: AWS’s profitability is partly inflated by Amazon’s own internal demand. The retail business is one of AWS’s largest customers. The logistics arm, the advertising platform, the device ecosystem — all of it runs on AWS. When you spin off the cloud unit, you don’t just separate two balance sheets. You rip out the demand floor that makes AWS look more profitable than it actually is.
AWS doesn’t stand on its own. It stands on top of the world’s largest shopping cart.
Think about what happens the day after a spinoff. AWS is now an independent company. It needs to negotiate a contract with Amazon Retail for compute, storage, and networking. Suddenly, the internal transfer pricing that made AWS’s margins look gorgeous gets exposed to market reality. Amazon Retail — now a separate entity with a fiduciary duty to its own shareholders — pushes back on pricing. Margins compress. The golden goose starts laying smaller eggs.
And it gets worse. Microsoft Azure and Google Cloud have been waiting for exactly this moment. Right now, they compete against an AWS that’s backed by the full financial weight of Amazon’s retail cash machine. That cross-subsidization lets AWS invest aggressively in custom silicon through Annapurna Labs, build data centers in places no pure-play cloud provider would touch, and undercut competitors on price during land-grab deals. Break that link, and AWS becomes just another hyperscaler — one with a smaller war chest than Microsoft.
The moment you spin off AWS, you don’t unlock value — you expose cost.
Now let’s talk about Annapurna Labs, because this is where the real strategic blindness shows. Amazon acquired Annapurna in 2015 for roughly $350 million — a rounding error. Today, that team designs the Graviton processors that power a huge chunk of AWS’s infrastructure. They build the Trainium chips for AI workloads. They’re the reason AWS can offer price-performance that Azure and Google can’t match.
Annapurna only works because it sits inside the loop. AWS tells Annapurna what the customers need. Amazon Retail tells AWS what the scale looks like. The data flows in a circle: retail demand feeds cloud scale, cloud scale feeds silicon investment, silicon investment feeds competitive advantage, competitive advantage feeds retail growth. Break any link in that chain, and the whole thing wobbles.
And here’s the twist nobody sees coming: a spun-off AWS would actually become more vulnerable to the hyperscaler war, not less. Right now, AWS benefits from Amazon’s diversification. When retail slows, cloud carries the company. When cloud faces pricing pressure, retail and advertising provide cushion. That’s not inefficiency — it’s a hedge fund’s dream portfolio wrapped inside a single corporate entity.
Investors who want the spinoff are asking Amazon to trade a competitive moat for a higher P/E ratio. That’s like a general trading his armor for a shinier sword. Sure, it looks great in the parade. But the first real battle will cut him in half.
Amazon isn’t a company. It’s a flywheel pretending to be three businesses.
The regulators want a breakup because they see monopoly. The shareholders want a breakup because they see undervalued assets. Both groups are looking at the same thing and missing the point entirely. Amazon’s power doesn’t come from the size of its parts. It comes from the friction between them — the way retail data trains the recommendation engine, the way cloud infrastructure subsidizes fulfillment logistics, the way custom silicon creates lock-in that no standalone cloud provider can replicate.
You don’t break up a flywheel. You either keep it spinning or you watch it stop.
So if you’re an investor salivating over the AWS spinoff narrative, ask yourself one question: would you rather own a piece of the most integrated competitive machine in business history, or would you rather own a cloud company that just lost its biggest customer, its silicon advantage, and its financial shock absorber — all on the same day?
Yeah. Thought so.
FAQ
Q: But wouldn't a spinoff let AWS trade at a higher multiple and unlock shareholder value?
A: Short-term, maybe. But that higher multiple would be built on inflated margins that disappear the moment Amazon Retail stops being a captive customer. You'd be paying a premium for a business that's about to get repriced.
Q: What does this mean for cloud customers?
A: If AWS gets spun off and loses its cross-subsidy cushion, expect pricing to get more aggressive — not in your favor. The cost advantages AWS currently passes through from Annapurna's custom silicon could erode without Amazon's integrated demand loop justifying the R&D spend.
Q: Isn't the real reason Amazon won't spin off AWS just that Bezos wants to keep power?
A: Power matters less than the flywheel. The integrated data loop between retail, cloud, and silicon is Amazon's deepest moat. Breaking it up wouldn't just reduce Bezos's influence — it would hand Microsoft and Google the competitive opening they've been waiting a decade for.