Supabase Compute: run agents next to your database

Supabase Compute, in private alpha, lets you run agents, background jobs, APIs and hours-long syncs inside your Supabase project, right next to your Postgres data, instead of on a second platform. Services run Node, Deno or any Dockerfile in two sizes (2 GB/1 vCPU or 4 GB/2 vCPU) and deploy with the Supabase CLI, so a coding agent can set them up for you. Fact or hype: real, but very early. It's waitlist-only, for evaluation rather than production, and only in the US West region (no UK or EU region), services can restart at any time, there's no autoscaling, GPU or persistent disk, and every service sees all your project's secrets. The product page also pitches agent sandboxes, scale-to-zero and autoscaling, but Supabase's own blog says those aren't in the alpha. It's free during the alpha; later pricing is per compute-hour and not yet published. The faster-queries example (2.9 ms against 65 ms) is a customer figure Supabase shared.

Why it mattersOne fewer platform for agent back ends is attractive, but don't put UK client data on it yet. Join the waitlist if you're curious; for real work today, Trigger.dev, Inngest, Modal, Railway, Render or Fly.io already do this.

How we knowCompany figures are labelled as their own claims; anything we couldn't verify is labelled as a report or a claim.