The Invisible Network Powering AWS with Matt Rehder

Our Take

Corey Quinn's Last Week in AWS sat down with AWS VP of Global Networking Matt Rehder for a genuinely interesting tour of the plumbing behind the cloud — flat network topology, custom silicon, SRD transport, the whole nerd-friendly stack. It's a rare, candid look at how a hyperscaler engineers around outages, and credit to both of them for pulling back the curtain on something most of the industry treats as a trade secret.

But notice the framing: Rehder talks about networks "as utilities" — always-on, invisible, just there when you need them. That's exactly the branding hyperscalers want the public to absorb. The problem is that AWS's network isn't actually a utility in the regulated sense — it doesn't answer to a public utility commission, and the actual utilities it leans on (regional power grids, water systems, local substations) absolutely do answer to someone, usually your city council or a state PUC that rarely gets asked hard questions before signing off on the deal. "Overbuild capacity" sounds great for engineers chasing five-nines uptime; it sounds a lot different when it's your community's power headroom being reserved for a hyperscaler's redundancy buffer.

None of this is a knock on the engineering — it's genuinely impressive work, and the AI-driven networking renaissance Rehder describes is real. It's just a reminder that the more invisible and resilient these systems become for AWS, the more visible their footprint gets on the ground: substations, transformers, land parcels, tax abatements. Check our facility map to see what's landing near you, and if a project's headed your way, our take-action page has the questions worth asking before your city signs anything.