Glossary · The AI Index

Hyperscaler

A hyperscaler is one of the world’s largest cloud-computing providers — typically Amazon Web Services, Microsoft Azure, Google Cloud, Meta, and Oracle Cloud Infrastructure — operating the data centers, networking, and AI accelerators that host most large-scale AI workloads.

How it works

Hyperscalers operate massive fleets of GPUs and custom AI accelerators (Google TPUs, AWS Trainium/Inferentia) inside purpose-built data centers, and rent compute, storage, and AI-platform services to customers on demand. Their scale lets them amortize land, power, and chip costs across millions of users — which is what makes frontier-model training and serving economically feasible.

Thank you for reading this post, don't forget to subscribe!

Why it matters

Hyperscalers are the gatekeepers of AI compute. Combined hyperscaler capex is projected to exceed $600 billion in 2026 — a roughly 36% jump over 2025 — with most of it going to AI infrastructure. For the full breakdown of who’s spending what, see our AI Investment & Funding Statistics 2026 and AI Infrastructure & Compute Statistics 2026.