Amazon Reports $15B Annualized AI Revenue Run-Rate

Amazon disclosed that AI services within Amazon Web Services (AWS) generated an annualized revenue run‑rate of more than $15 billion in the first quarter of 2026. CEO Andy Jassy revealed the figure in the company’s April shareholder letter and used it to justify a $200 billion capital expenditure plan for 2026 focused mainly on AI.

AWS’s AI revenue run‑rate equals roughly 10% of the $142 billion AWS revenue run‑rate cited in the same disclosure. Jassy framed the $200 billion capex as forward‑looking infrastructure investment — with monetization expected in 2027–2028 — and said a substantial portion already has customer commitments. The number sits alongside Microsoft’s earlier disclosure of a roughly $13 billion AI run‑rate, signaling that hyperscalers are converting AI demand into high‑growth cloud businesses.

The $15 billion figure is derived from first‑quarter performance and is the first time Amazon has published an explicit AI revenue run‑rate for AWS. The shareholder letter links the capex plan directly to AI capacity buildout and custom hardware investments (implicit in the capex focus). Market reaction was immediate: Amazon shares rose about 4.6% after the announcement. Investment managers quoted in coverage described the run‑rate as validation that AWS can monetize the AI wave at scale.

This disclosure materially de‑risks the AI infrastructure market narrative for engineers and architects: demand for managed AI services and cloud inference/training capacity is now a multi‑billion‑dollar revenue stream at AWS. Expect continued prioritization of custom accelerators, colocated storage/network upgrades, pricing experiments, and more managed model services from AWS as the company seeks to convert 2026 capex into 2027–2028 revenue. For ML engineers and platform teams, the announcement signals expanding cloud capacity and sustained vendor investment rather than a speculative, short‑term cycle.

how AWS phases the announced capex into capacity (chips, regions, racks), product roadmap changes for managed model services and custom chips, and competitive pricing or commitment programs in response to Microsoft’s similar momentum.