Amazon Wouldn’t Be Doing What It’s Doing if AI Capex Was Going to Fail
Andy Jassy just committed roughly $200 billion in capital expenditures for 2026, and the logic behind that number reveals something most investors are missing about where AWS demand is actually coming from.
This post may contain links from our sponsors and affiliates, and Flywheel Publishing may receive compensation for actions taken through them.
I keep hitting the buy button on Amazon (NASDAQ:AMZN | AMZN Price Prediction) because Amazon would not be unleashing the AI capex it is unleashing if there was a reasonable chance it was going to fail. That is the whole thesis in one sentence. When a management team with a two-decade record of long-duration bets tells shareholders it plans to spend about $200 billion in capital expenditures across 2026, that reflects a company reading its own order book and building against it.
What the Order Book Actually Says
The order book is the part I cannot ignore. AWS backlog sat at $496 billion, growing triple digits year over year, on a segment already running at a $169 billion annualized revenue run rate. Andy Jassy told investors the lion’s share of AWS capacity planned for 2027 is largely reserved, with “quite a bit of capacity” already reserved for 2028. Customers are pre-committing years of demand. Amazon is pouring concrete against contracts.
The economics of that concrete are what convince me to keep adding. Jassy walked through it plainly: data centers have 30-plus year useful lives, servers break even in a little less than three years and then run for at least five to six years, with most AI capacity contracted for at least five-year terms. That is a compounding annuity dressed up as a capex line.
Numbers That Earned My Conviction
Q2 FY2026 gave me three receipts. AWS grew 37% year over year, the fastest AWS growth in 18 quarters, at a 39.4% operating margin. Operating income across the company rose 43% year over year to $27.46 billion. And the AI and custom chip businesses each cleared a run rate of more than $25 billion, both growing at triple-digit rates. Trainium2 is fully subscribed, and Anthropic committed up to 5 GW while OpenAI committed about 2 GW starting 2027. Balance sheet strength backs the build: debt/equity of 0.37, net debt/EBITDA of 0.45, interest coverage of 35.17.
Why AMZN Instead of the Other Two Hyperscalers
The reader reaches for Microsoft (NASDAQ:MSFT) or Alphabet (NASDAQ:GOOGL). I own the cloud franchise with the fastest listed acceleration and a P/E of 35 attached to a business Jassy believes could be a trillion-dollar annual revenue business for us in time. Amazon also owns something the other two do not: a vertically integrated chip stack where Graviton is used by 98% of Amazon’s top 1,000 EC2 customers and management is actively exploring selling Trainium chips separately from AWS cloud services. That is optionality I am not paying for.
Risk I Refuse to Wave Away
Free cash flow already turned negative at -$7.6 billion TTM, with Q2 capex of $54.21 billion, up 68.4% YoY. If demand cracks, that spend becomes a stranded asset problem. I sit with that risk because the contracted backlog, five-year customer terms, and 30-year data-center life all say the demand is already booked. Jassy’s own words: “We’re enthusiastic about the ROIC equation, even with heavy CapEx the next few years.”
Why the Buy Button Stays Active
Jassy says 85% of global IT spend is still on-premises and that equation flips over the next 10 to 20 years. My retirement horizon is exactly that long, and I plan to own the company pouring the foundation while everyone else argues about the paint.
Contact [email protected] for any questions or corrections.
Alex Sirois is a financial writer with experience spanning both retail and institutional investing. He has written for InvestorPlace and held roles at BNY Mellon and Bernstein, giving him a perspective that bridges Main Street portfolios and Wall Street analysis.
Alex holds an MBA from George Washington University and has built his career across multiple industries, including e-commerce, education, and translation — a breadth of experience that informs how he breaks down complex financial topics for everyday investors. His writing is conversational, actionable, and grounded in long-term, buy-and-hold investing principles.
At 247 Wall St., Alex focuses on delivering analysis that is both accessible and useful, with a clear emphasis on helping readers make more informed decisions with their money.
