Fly.io
Run full-stack apps and databases on global microVMs.
Description
Fly.io is a public cloud platform built by Kurt Mackey, Michael Dwan, Jérôme Gravel-Niquet, and Thomas Ptacek, founded in 2017 and refined through Y Combinator's Winter 2020 batch. The platform deploys application servers, databases, and GPU workloads across 30+ regions on hardware-virtualized microVMs called Fly Machines. Headquartered in Chicago and operated remote-first, Fly.io targets the latency gap that traditional CDNs leave open: dynamic application code running close to users instead of just static assets.
Key Capabilities:
Hardware-virtualized microVMs with millisecond-class boot times
Docker-based deployment through the flyctl CLI and Machines API
Global edge deployment across 30+ regions with anycast IP routing
Managed Postgres with HA clusters, automatic failover, and pgvector support
LiteFS for distributed SQLite replication across regions
Tigris S3-compatible global object storage
Persistent NVMe volumes attached to Machines
WireGuard private networking and TLS termination via Fly-Proxy
GPU-accelerated workloads for AI inference and training
Autostop and autostart for scale-to-zero cost efficiency
Hardware-isolated security model with end-to-end encryption
Phoenix.new remote AI runtime for Elixir application scaffolding
Alternative tools
- Modal
Serverless GPU platform for AI inference, training, and batch jobs.
- Baseten
Production AI inference platform for custom and open-source models.
- DeployLens
Real-time visibility into your GitHub deployment pipeline.
