Amazon's $200 Billion Gamble: Why I'm Buying the Stock Everyone's Selling

By Glenn Ford @ GNG_Research | February 16, 2026 Amazon (AMZN) just told Wall Street it plans to spend $200 billion this year on capital expenditures. The stock has shed nearly 20% from its November peak, with over 10% of that damage coming in the single week after earnings. Analysts are calling it reckless. Prediction markets collapsed within minutes of the announcement. And I started buying. Not because I think the market is wrong about the risk. The risk is real. Two hundred billion dollars exceeds the entire annual economic output of countries like Ukraine or Kuwait. It is more than what most countries spend on their entire military. It is, by any historical measure, one of the largest single-year corporate investment commitments ever made. But risk and opportunity are two sides of the same blueprint, and right now, Amazon is drawing up plans that could define the next decade of computing. Think of it this way. In 1869, the transcontinental railroad was finished, connecting the coasts of a vast continent. The skeptics focused on the cost of every mile of track. The visionaries focused on what moved across those rails. Amazon is building the digital railroad of the AI era, and the market is fixated on the price of the steel. The Three-Engine Machine at a Crossroads Amazon runs three businesses that most companies would kill to own just one of. AWS generated $35.6 billion in Q4 revenue alone, growing 24%, its fastest pace in 13 quarters. The advertising division pulled in $21.3 billion, growing 22%. And the core retail operation, which skeptics love to dismiss, pushed total company revenue to $213.4 billion for the quarter and $716.9 billion for the full year. Operating income hit $25 billion in Q4, with AWS alone contributing $12.5 billion at a 35% margin. Those are not the numbers of a company in trouble. Those are the numbers of a company whose stock is being punished for investing too aggressively in its own future. The custom silicon story deserves special attention. Amazon's Trainium and Graviton chips now run at a combined annual revenue rate exceeding $10 billion, with triple-digit year-over-year growth. Trainium2 is fully subscribed with 1.4 million chips deployed. Amazon Bedrock, the managed AI platform, is used by over 100,000 companies, and customer spending grew 60% quarter over quarter. These are not science projects. These are production workloads generating real revenue at scale. Why the Market Is Punishing a Winner The Feb 5 earnings call dropped three numbers that sent traders scrambling. First, EPS of $1.95 missed the $1.97 consensus by two pennies. Second, Q1 2026 operating income guidance of $16.5 to $21.5 billion came in below the $22.2 billion Street estimate. Third, and most critically, the $200 billion capex target for 2026 obliterated the $146.6 billion analysts had modeled. That last number changed the math for every model on Wall Street. Free cash flow for the trailing twelve months was just $11.2 billion, down from $38.2 billion a year earlier, driven entirely by the jump in infrastructure spending. The market looked at that trajectory, extrapolated the $200 billion commitment, and concluded that meaningful margin expansion was dead for the foreseeable future. Here is where I diverge from the consensus panic. The market is treating all capex as if it were maintenance spending, the cost of keeping the lights on. My framework separates maintenance capex (roughly 35% of total spend) from growth investment that should earn returns above the cost of capital. When I normalize free cash flow using that split, the picture changes dramatically. Operating cash flow of $139.5 billion minus an estimated $46 billion in maintenance capex yields a normalized free cash flow power closer to $93 billion, not the $11.2 billion headline number. That normalization is not wishful thinking. It is the difference between valuing a company building a railroad and valuing a company maintaining one. Va