Amazon CEO: Inference Demand Surges, $220B Capex
Amazon.com Inc.’s cloud division, Amazon Web Services, is witnessing a shift from AI model training to inference, the stage where trained models are used to respond to user prompts. AWS Chief Executive Officer Matt Garman explained on Bloomberg Technology television on Monday that while large training clusters remain in use, “more and more companies are integrating that inference into their workloads.”
The company disclosed that it will spend $220 billion in capital expenditures for fiscal year 2026, an increase from its earlier forecast of $200 billion. The uplift reflects rising costs for memory and other components required for data‑center expansion. Garman did not provide a spending outlook for the following year but described Amazon’s artificial‑intelligence business potential as “just massive.”
To align investment with demand, Amazon is obtaining five‑year commitments from customers, enabling it to project usage for new data‑center capacity. Garman emphasized that “today, demand still significantly outstrips supply and we’re trying to build and invest to keep up with what customers are asking for.”
Amazon’s share price has risen, lifting the company’s market valuation above $3 trillion. In the most recent quarter, cloud‑computing revenue recorded growth for the fifth consecutive quarter.
---