Amazon to spend $220B on AI infrastructure in 2026—still won't meet demand

AWS has emerged as a powerful growth engine within Amazon, now operating at a $169 billion annualized revenue run rate and delivering its fastest growth in 18 quarters. The cloud unit's Q2 revenue of $42.2 billion, up 37% year-over-year, reflects surging demand for cloud and AI services, evidenced by a backlog that swelled to $496 billion. This backlog—representing customer agreements for future revenue—underscores persistent, pent-up demand for cloud infrastructure that Amazon is racing to fulfill.

However, the company faces a structural constraint: even as it dramatically escalates capital spending from $200 billion to $220 billion in 2026, CEO Jassy explicitly stated that Amazon will still lack sufficient capacity to meet demand in 2026 and expects this shortage to persist into 2027. This capacity gap reflects broader industry dynamics in which hyperscalers and cloud providers are committing more than $800 billion across the sector to data center and AI infrastructure buildout, yet demand for these resources continues to outpace supply. The earnings call reveals an industry in which infrastructure scarcity, not demand creation, is the binding constraint—a dynamic that may persist for years despite unprecedented capital deployment.