Amazon sells $25B in bonds to fund record AI infrastructure push

Amazon is doubling down on a vertically integrated AI strategy, using debt to finance the hardware and data centers required to stay ahead of Microsoft and Google. For the streaming and cloud ecosystem, this massive capital outlay ensures the infrastructure for next-generation generative AI and recommendation engines remains robust, even as it shifts margins toward proprietary silicon like Trainium. The move signals that market leaders are prioritizing compute capacity over short-term profitability, betting that the lack of infrastructure is a greater risk than the cost of debt. Watch for whether AWS can maintain its 28% growth rate as this new capacity comes online through 2026.

The borrowing spree underscores a broader 'arms race' among hyperscalers. For 2026, the combined capital expenditure of Amazon, Alphabet, Microsoft, and Meta is expected to hit approximately $725 billion, a 77% increase over 2025 levels, per ValueAddVC in June 2026. Microsoft alone signaled its fiscal 2026 spending would exceed $88.2 billion, while Alphabet guided between $175 billion and $185 billion, according to Investing.com in February 2026. This massive investment has faced scrutiny; Amazon shares previously fell more than 10% following the initial $200 billion capex disclosure due to concerns over long-term returns on invested capital, as reported by MLQ News in February 2026.

Simultaneously, Amazon is pivoting from a consumer of chips to a major producer. CEO Andy Jassy revealed in April 2026 that AWS custom silicon—including the Graviton CPU and Trainium AI accelerator—has reached an annual revenue run rate exceeding $20 billion, growing at triple-digit percentages, per Investing.com. To support this growth, Amazon secured a $100 billion infrastructure commitment from OpenAI and a similar arrangement with Anthropic, involving the use of Trainium chips through their fourth generation, according to CRN in July 2026. The company’s ability to sell its own hardware to major AI model builders is becoming a critical offset to its record infrastructure spending.