Vercel Ship 2025 recap

By Steven Van · 2025-06-26

The June event introduced AI Gateway, Active CPU pricing, Vercel Sandbox, Rolling Releases, Microfrontends, and BotID bot detection.

At Vercel Ship 2025, held in New York in June, Vercel announced a set of platform updates aimed at building AI-powered apps and agents.

AI Gateway gives a single endpoint for models from providers including OpenAI, Anthropic, Google, and xAI, with usage-based billing at provider list prices, bring-your-own-key support, and provider switching with one line of code.
Active CPU pricing for Fluid compute now charges only for the time code is actually executing, with idle wait time billed as Provisioned Memory at 1/11th the rate. The default execution time for all projects rises to 300 seconds, up from 60 to 90 seconds previously.
Vercel Sandbox is a new isolated, ephemeral environment for running untrusted or AI-generated code, supporting Node.js and Python, scaling to hundreds of concurrent environments, and allowing execution times up to 45 minutes. It works as a standalone SDK outside Vercel as well.
Rolling Releases let teams roll out a deployment to a subset of users first, with monitoring of metrics like Time to First Byte and error rates, and the ability to pause or abort a rollout. Updates propagate globally in under 300ms.
Microfrontends lets teams split large applications into independently deployable units, each built and deployed on its own, while Vercel handles integration, routing, and unifies everything in one dashboard.
BotID, built with Kasada, adds invisible bot detection for routes like checkouts, logins, signups, and APIs, aimed at bots that execute JavaScript and solve CAPTCHAs.

The full list of announcements and keynote video are on the Ship 2025 recap page.

Vercel

The platform for frontend developers — deploy, preview, and scale web apps and AI agents with zero config.