Forty-four analysts rate Amazon a Strong Buy. The average price target implies 36% upside. Jeff Bezos sold $5.7 billion in Amazon shares in 2025. Follow the money. Not the narrative.
The consensus view on Amazon right now is seductively simple: great company, temporarily depressed stock, obvious entry point. A 22% drawdown from the highs. A forward P/E near a decade low. AWS growing 24%. What's not to love?
Plenty. And nobody is talking about it.
The Depreciation Tsunami Nobody Is Modeling
The bull case focuses almost entirely on the $244 billion AWS backlog and the idea that Amazon's $200 billion capex plan is a rational response to insatiable demand. Maybe. But here's what happens after you spend $200 billion on physical assets: you depreciate them. And that depreciation hits your income statement for years.
Amazon's depreciation and amortization expense already surged 24.5% year-over-year to $65.8 billion in 2025. That number is going dramatically higher. When you layer $200 billion in new capital expenditures on top of an existing asset base that's already generating nearly $66 billion in annual depreciation, you are looking at a depreciation expense that could approach $80–90 billion within 18 months. That flows straight through operating income.
And here's the part that should make you uncomfortable. In early 2025, Amazon actually shortened the useful life estimate on a subset of its servers and networking equipment from six years to five, citing the "increased pace of technology development, particularly in the area of artificial intelligence." That change increased depreciation expense by roughly $1 billion in the first half of 2025 alone. Think about that: the company is telling you, in its own filings, that AI hardware is obsolescing faster than expected. Then it's spending $200 billion on more of it.
The 26x forward P/E looks cheap until you realize the "E" in that equation is about to get compressed by a wall of depreciation expense that hasn't fully hit yet. Consensus 2026 EPS of $7.72 assumes 7.7% earnings growth. That estimate has nowhere to go but down as the depreciation from this capex cycle flows through.
The $700 Billion Arms Race With a 42% Coverage Ratio
Zoom out from Amazon for a moment and look at the industry. The four largest hyperscalers plan to spend close to $700 billion in combined capex this year. That figure exceeds the GDP of most countries. And the revenue being generated by AI products? One analysis pegged total trackable AI revenue across the major players at roughly $223 billion against $527 billion in hyperscaler spend, a coverage ratio of just 42%.
For context, when cloud computing was at a comparable stage of its adoption curve around 2011, the spending-to-revenue ratio was approximately 2.4-to-1. The current AI ratio is running above 10-to-1. That gap is where bubbles live. The supply side is scaling at a pace never seen in corporate history, while the demand side, the actual end-user businesses supposed to justify all this infrastructure, remains stuck in what the industry diplomatically calls "pilot purgatory."
Bank of America's credit strategists have flagged that five of the Magnificent Seven are now pushing capex to 94% of operating cash flows. RBC's Lori Calvasina told clients that "risk of AI overspend/overhype" has moved from healthy rotation to "outright derisking." Evercore ISI called the current free cash flow trajectory a "red flag" for stock valuations. When everyone is bullish, I get nervous. When the credit analysts start flagging things equity analysts are ignoring, I get very nervous.
The Signals the Stock Price Is Sending
Amazon filed a shelf registration statement with the SEC in February, allowing it to issue debt securities, common stock, preferred stock, warrants, and other instruments. The filing language is boilerplate. The timing is not. A company about to go negative $17 to $28 billion in free cash flow, which just announced $200 billion in capex, is telling you in black and white that it may need to tap the debt or equity markets. Alphabet already quadrupled its long-term debt to $46.5 billion in 2025 and held a $25 billion bond sale. Amazon issued nearly $15 billion in senior notes in November 2025, with maturities stretching to 2065.
Meanwhile, over the last 90 days, Amazon insiders executed 37 transactions totaling over $56 million in sales. Zero insider buys. Not one. The CEO of AWS, Matt Garman, is selling. The CEO of Worldwide Stores, Doug Herrington, is selling. Bezos himself unloaded 25 million shares for $5.7 billion in 2025. I understand that insiders sell for many reasons. But when nobody on the inside is buying a stock that's down 22% from its high, at what the bulls insist is a generational entry point? That should at least give you pause.
The last time Amazon had a nine-day consecutive losing streak was July 2006. That's almost 20 years. The market is not punishing Amazon for bad earnings. The earnings were fine. It's punishing Amazon for making a bet so large that even the bulls can't model the return with confidence. As Albert Bridge Capital's Drew Dickson put it, "the market expects [capex] to translate into revenue growth in a time horizon that makes little sense."
Amazon may very well emerge from this capex cycle as the dominant AI infrastructure company on the planet. It has the balance sheet, the backlog, and the talent to make it work. I'm not arguing the company is broken. I'm arguing the stock is not discounting the actual risk. A forward P/E near decade lows can go lower. It went to single digits in 2008. Capital intensity ratios at 45–57% of revenue look more like a utility company than a tech platform that once commanded premium multiples precisely because it was asset-light. That model is dead.
When a company is spending more than it earns, borrowing to fill the gap, accelerating depreciation on its own hardware, and its insiders are net sellers, the correct response is not to back up the truck. It's to size your position like the outcome is uncertain. Because it is.
