FLY.IO

Kurt Mackey had run ops teams and been frustrated by the complexity of running containers globally. The core problem: running an application close to your users (for low latency) meant either spending enormous time managing multi-region Kubernetes clusters or paying cloud provider premiums.

Fly.io's approach is different — it runs applications at the edge, across 30+ regions globally, using a custom orchestration layer built on Firecracker micro-VMs. The deployment model is simple: wrap your app in a Docker container, point it at Fly, and it deploys globally in seconds.

For latency-sensitive applications, the performance difference versus a single-region deployment is dramatic.

Fly.io charges developers for compute (VM time), memory, bandwidth, and persistent storage. Pricing is transparent and usage-based — you pay for what runs.

The model is deliberately simple: no egress fees from Fly.io's network (a significant cost driver on AWS), no complex instance pricing tiers.

Fly.io app deployments (Docker-based, 30+ regions), Fly Machines (Firecracker micro-VMs), Fly Volumes (persistent storage), Fly Postgres, Fly Redis

Fly.io grew through developer word-of-mouth — particularly among builders running Elixir/Phoenix and Go applications that benefit most from its distributed model. Its transparent pricing (no egress fees, no hidden costs) became a genuine differentiator as AWS egress charges became a community-wide frustration.

Tiger Global and a16z investments brought growth capital without fundamentally changing the product philosophy.

Fly.io competes against Railway, Render, Heroku, and the major cloud providers on the simple deployment side. On the distributed compute side it competes against Cloudflare Workers and edge-specific platforms.

The market for developer-friendly cloud hosting is crowded, and Fly.io's edge architecture adds complexity that not every application needs.

Base10 Partners, Tiger Global, Andreessen Horowitz