Let’s begin with AWS’s new networking philosophy, which focuses on making network connectivity nearly invisible to users and administrators alike. For AWS, networking needs to be as reliable as flipping a switch—it simply works, and no one notices unless it fails. To meet this lofty goal, AWS spent the past decade moving away from traditional, proprietary network hardware and has built a unified, custom stack that spans everything from silicon to software.
At the heart of this innovation is the decision to use a single switching application-specific integrated circuit (ASIC)—the silicon “brain” at the center of every network switch—across its aggregation, core, and border networks. Rather than the old industry practice of mixing and matching hardware from various vendors (each with its own firmware, bugs, and scaling challenges), AWS has network engineering and operations teams focus on a single, consistent foundation. This not only simplifies procurement and troubleshooting but also enables scalable, reproducible deployments that surpass the speed of conventional enterprise or cloud architectures.
All these switches run NetOS, a Linux-based operating system that AWS built to optimize security, automation, and rapid patching. If a bug appears, AWS can fix it everywhere, instantly; there’s no waiting for vendor patches and no risk that yesterday’s firmware problems could spiral into tomorrow’s outage.
The numbers are staggering. The current AWS switch handles 51.2 terabits per second (Tbps) across 64 ports, each operating at 800Gbps. But the pace of innovation is accelerating: AWS’s next-generation switch, coming soon, will reach 102.4Tbps, with ports running at 1.6Tbps. Combined, AWS’s network comprises roughly 2 million devices, 50 million to 60 million optical links, and more than 20 million kilometers of fiber—enough to circle the moon 25 times.
