Every hyperscaler is built on circular dependencies

Doron Grinstein, CEO of Control Plane and former chief architect of VMware’s cloud services platform, on why the October 2025 AWS us-east-1 outage wasn’t a one-off.

A latent bug in DNS cascaded and took down tens of thousands of clients. Doron’s point is that this can happen at any hyperscaler, because their core services depend on each other in a loop: compute depends on identity, identity on networking, networking on metadata, metadata on a database, and the database on compute.

In his words: “This is not unique, it’s not a one-time deal, and it can happen to any hyperscaler because of how they’re architected with circular dependencies.”

From our webinar recreating the us-east-1 outage live.

Every major AWS, Azure, Google Cloud and Cloudflare outage since 2020, with each provider’s own timeline and root cause: https://controlplane.com/cloud-outage-tracker
Control Plane: https://controlplane.com