Amazon Raises 2026 AI Spending Forecast to $220 Billion

Following a strong second-quarter earnings report, Amazon.com increased its projected capital expenditure for 2026 to approximately $220 billion, up from an initial estimate of $200 billion. The spending surge is driven by the rising cost of memory chips and the need to expand AI infrastructure and Amazon Web Services (AWS) data centers. AWS reported a 36.7% year-over-year revenue increase to $42.2 billion, marking its strongest performance in over four years.

CEO Andy Jassy warned that AI capacity will likely remain constrained through 2027, with demand already extending into 2028. To manage this growth, Amazon is securing long-term customer commitments, including five-year agreements. The company is focusing on custom silicon, noting that its AI and custom chip businesses have each surpassed a $25 billion annualized revenue run rate. While Amazon continues its partnership with Nvidia, it is scaling its Trainium and Graviton chips to offer customers cost savings of 20% to 30% on inference workloads.

This investment is part of a broader industry trend. Morgan Stanley predicts cloud capital spending will reach $1.4 trillion by 2027, while the combined backlogs of the top four hyperscalers have reached $2.3 trillion. Amazon's AWS backlog specifically reached $496 billion, a triple-digit year-over-year increase. Other major players like Alphabet Inc. also raised 2026 guidance, while Microsoft maintained its existing investment plan.