Serverless and Edge Functions

How each cloud/hosting platform runs short-lived, on-demand functions — only covering platforms that actually offer this.

Render, Fly.io, and Heroku are intentionally omitted here — none of the three offer a dedicated serverless/edge-function product; they run long-lived containers or VMs instead. See Compute Model for how those three actually run code.

AWS

Lambda, the service that popularized the “serverless function” model — functions triggered by HTTP requests, queues, schedules, or almost any other AWS event source, billed per invocation and execution time.

GCP

Cloud Functions (and increasingly Cloud Run, which also supports request-triggered, scale-to-zero containers) for event-driven and HTTP-triggered serverless workloads.

Azure

Azure Functions, triggered by HTTP requests, timers, or events from other Azure services, with a consumption plan that scales to zero like its AWS and GCP counterparts.

Vercel

Serverless Functions (Node.js/Python/etc., run per-region) and Edge Functions (lighter-weight, V8-isolate-based, run at the edge), automatically chosen based on how a project’s API routes are configured.

Cloudflare

Workers are the platform’s core offering, not an add-on — V8-isolate-based functions with near-zero cold start that run at the edge across hundreds of cities.

Netlify

Functions (AWS Lambda-backed) and Edge Functions (Deno-based, running at the edge), mirroring Vercel’s two-tier model of regional versus edge execution.

DigitalOcean

DigitalOcean Functions, based on the open-source Apache OpenWhisk project (via DigitalOcean’s acquisition of Nimbella), supporting Node.js, Python, Go, PHP, and other runtimes.