Serverless

Serverless is a cloud model in which you deploy code and the provider handles provisioning, scaling, and patching of the machines that run it. Servers still exist; you simply never manage them. The defining traits are event-driven execution, automatic scaling from zero to many concurrent instances, and billing by request and execution time rather than by reserved capacity. AWS Lambda established the category, and Google Cloud Functions, Azure Functions, Vercel Functions, and Cloudflare Workers followed, with managed databases and queues extending the same idea beyond compute. The fit is bursty or unpredictable traffic, webhook handlers, scheduled jobs, and API endpoints for small teams that would rather not run infrastructure. The constraints are real. Cold starts add latency when an idle function is first invoked, execution time is capped, functions are stateless so session data must live elsewhere, and local development differs from production. Cost inverts at high steady volume, where a reserved container is cheaper than per-invocation pricing. Long-running AI inference sits awkwardly here for the same reasons, which is why many teams pair serverless APIs with dedicated GPU services.