Vercel Deploys Failing at “Deploying outputs…” (March 2026) – What’s Going On?

Today (March 2, 2026), a lot of Next.js apps on Vercel are suddenly failing to deploy with this message, even though the build itself is green:

Traced Next.js server files in: XXms
Created all serverless functions in: XXXms
Collected static files (public/, static/, .next/static): XXms
Build Completed in /vercel/output [Xs]
Deploying outputs...
Error: We encountered an internal error. Please try again.

For many teams, this broke pipelines that had been stable for weeks or months. Builds succeed, artifacts are generated, and then everything dies in the “Deploying outputs…” phase.

It’s (probably) not your code From the active community threads and incident reports, this looks like a Vercel-side infrastructure problem, not a bug in your Next.js project:

• Multiple regions (iad1, sfo1, sin1, dxb1) are affected.

• Previously working commits now fail with the same internal error and no additional stack trace.

• Deployments that pass npm install and npm run build locally still fail in Vercel at the “Deploying outputs” step.

If your logs show a clean build and only fail once Vercel starts “Deploying outputs…”, you’re almost certainly hitting the current platform issue.

Quick sanity checks before you blame Vercel You should still rule out the usual suspects, so you’re not masking a real problem:

• Build locally: Run npm run build or pnpm build locally/CI. If that fails, fix those errors first.

• Check install logs for package issues: For example, the supabase CLI package (not @supabase/supabase-js) currently has a failing postinstall.js in some environments. If you see something like: ​

npm error path /vercel/path0/node_modules/supabase
npm error command sh -c node scripts/postinstall.js

then you should remove supabase from dependencies/devDependencies for your Vercel build and keep it as a local-only tool.

• Watch for Edge/middleware errors: If your middleware or Edge functions reference unsupported Node APIs, you’d see errors like “Edge Function middleware is referencing unsupported modules”. That’s different from this silent “Deploying outputs” internal error.

If all of that looks clean and the log only dies at “Deploying outputs…”, this is almost certainly on Vercel.

Current workarounds people are trying None of these are guaranteed, but they have helped some teams ship while the incident is ongoing:

1. Switch your Functions region In Vercel Project Settings → Functions, change your region (for example from dxb1 or sin1 to fra1, iad1, or sfo1), then redeploy. Some regions have higher failure rates during incidents.

2. Clear build cache / force no cache

• Use the dashboard: “Redeploy” → “Clear build cache & redeploy”.

• Or set a temporary env var like VERCEL_FORCE_NO_BUILD_CACHE=1 and red eploy, then remove it later. ​

1. Retry after a short delay There are multiple reports of deployments randomly succeeding later with no code changes once the underlying infra issue is mitigated. ​

How I’m approaching it as a maintainer

• I now treat this specific pattern (clean build, failure only at “Deploying outputs…”) as a hosting-layer incident, not a personal failure.

• I validate with local/CI builds, keep my codebase intact, and avoid desperate, breaking refactors just to “appease” Vercel.

• I communicate to stakeholders: “Our app builds and passes tests; the hosting provider is having deployment issues.”

I keep an eye on:

• Vercel status: https://www.vercel-status.com

• Community threads like “Vercel deployment failu re with internal error during deploying outputs phase”.

Takeaways

• If you’re seeing this today, you’re not alone.

• Don’t nuke your architecture or delete half your app because of this one log line.

• Use region switching and cache clearing as temporary tools, and wait for a proper fix from Vercel.

If you’ve found any other reliable workaround, drop it in the comments so others don’t have to go through the same “it must be my code” spiral.