Compare / Vercel
Watasu vs Vercel for full-stack apps with a real backend
Watasu and Vercel both deploy your code from Git, spin up a preview environment for every pull request, issue TLS certificates for your domains and show you logs without a single server to manage. Under the hood they are built for different jobs.
Vercel started as the home of Next.js and a global frontend CDN, and now positions itself as infrastructure for apps and agents. Its core is still serverless: server code runs as functions on Fluid compute on AWS. Over the past year it has added Python backends, queues, workflows, sandboxes and always-on Daemons, several of them still in beta. Watasu is a European platform as a service. It runs your whole product, from the frontend and APIs to background workers, databases and AI sandboxes, as long-running processes in German data centers, on one prepaid balance in euros.
This page compares the two hosting platforms on infrastructure, data location, services, pricing and developer workflow, and then shows how to move a Vercel project to Watasu. If you are looking for a Vercel alternative because your app has outgrown serverless functions, start with the table.
Who should choose which
Choose Watasu if
- Your product has a real backend: APIs that do work, queues, scheduled jobs, WebSocket connections, a database or AI agents.
- Your users or customers are in Europe and ask where their data lives. Everything runs in Germany and the company is not US-based.
- You want one bill in euros for compute, databases, bandwidth and builds, with no per-developer seats.
Choose Vercel if
- Your product is mostly a frontend: a marketing site or content-heavy Next.js app with visitors on every continent who benefit from edge caching.
- Your API routes are short and spiky, mostly idle, and fit inside the $20 monthly credit of Vercel Pro.
- Your team relies on Vercel-specific products such as v0 or the AI Gateway.
What Watasu and Vercel have in common
- Git-based deploys. Connect a GitHub repository and every
git pushto a branch builds and deploys automatically. - Preview environments. Each pull request gets its own URL to review the change before it merges. On Watasu, review apps also get their own databases and other add-ons.
- Domains and automatic SSL. Custom, apex and wildcard domains, with SSL certificates issued and renewed for you.
- Environment variables and built-in observability. Config injected at runtime, logs out of the box, deeper metrics and tracing available.
- A CLI for everything. Both platforms can be driven entirely from the terminal.
Watasu vs Vercel at a glance
The rows that usually decide. Vercel figures are for the Pro plan; Watasu figures are for Standard and Dynamic pods. Shared features are listed above, not repeated here.
| Row | Watasu Standard / Dynamic pods | Vercel Pro plan |
|---|---|---|
| How your code runsThe execution model shapes everything else: limits, state, pricing. | Long-running containers: web, worker, private TCP and real-time processes | Serverless functions on Fluid compute (AWS); containers and always-on workers in beta |
| Execution timeHow long one request or job may run before the platform stops it. | Processes run continuously | 300 s by default, up to 800 s on Pro, 1,800 s in beta |
| CPU and memory per instanceCeiling for one replica of a service. | Up to 16 vCPU / 30 GiB | Up to 2 vCPU / 4 GB on Pro |
| WebSocketsPersistent connections for chat, live updates, streaming agents. | Supported on every web process | Public beta; connections close at the function's max duration |
| WebRTC and UDPNeeded for voice and video media, not just signalling. | Dedicated TURN gateway and public UDP per real-time process | Not available |
| Background workersQueue consumers, schedulers, long imports that run without an HTTP request. | Worker processes, scaled independently | Daemons, public beta: no inbound traffic, production only, no autoscaling |
| DatabasesWho runs them, and whether they sit on the same bill and in the same region. | PostgreSQL, Valkey, ClickHouse, Qdrant, Redpanda and S3-compatible storage, run by Watasu | Through Marketplace partners such as Neon, Supabase, AWS Aurora and Upstash, plus Vercel Blob |
| Where it runsData location and the company's jurisdiction. | Data centers in Germany across several zones; all customer data stays in the EU. Watasu LTD | AWS; functions default to Washington, D.C. (iad1), EU regions available; global CDN. Vercel Inc., United States. CLOUD Act implications |
| Pricing modelWhat the monthly bill is made of. | Prepaid balance in EUR: fixed price per replica, or measured usage with a monthly ceiling | $20 per month per developer seat plus usage across CPU, memory, invocations, builds and CDN tiers |
| Seats, bandwidth, buildsThe line items that grow with the team and the traffic, not with the code. | No per-seat fees; normal app bandwidth included; builds €0.50/hour by running seconds | $20 per developer seat per month; 1 TB included, then Flat Rate CDN tiers from $20 per month; builds billed per minute |
| Also worth knowing | ||
| AI sandboxesIsolated environments for running generated or untrusted code. | Isolated sandboxes with per-second billing, in the EU | Vercel Sandbox, billed by Active CPU |
| DockerWhether you can ship your own image. | Dockerfile or Cloud Native Buildpacks | Container support in beta |
Running something that has to stay up? Standard pods start at €15 per month for 1 vCPU and 2 GiB, with no per-seat fees and normal app bandwidth included. App builds cost €0.50/hour by running seconds.
See pricingPricing
Watasu app builds have limits of 16 vCPU and 30 GiB RAM and cost €0.50/hour by actual running seconds, including failed and canceled builds. Queue time and image-only pipeline promotions have no build charge. Before execution, a temporary hold of up to €0.50 funds at most one hour, or a shorter deadline from a smaller available balance; unused funds return after the build stops. Enough available balance for one second is required, with no saved-card requirement or free monthly build allowance. App runtime examples below exclude build charges unless a build row is shown.
Vercel Pro costs $20 per month per developer seat and includes $20 of usage credit. Beyond that, usage is billed on several meters: Active CPU, provisioned memory, invocations, build minutes and CDN capacity, with compute rates that depend on the region. New teams get a default $200 spend alert and can choose to pause projects when they reach it. Bandwidth overages above 1 TB move you to Flat Rate CDN tiers.
Watasu has no platform fee and no seats. You top up a balance in euros (from €5) and usage draws it down. Each process picks one of two billing models:
- Standard: a fixed monthly price per replica for sustained load, for example €15 for 1 vCPU and 2 GiB.
- Dynamic: measured CPU and memory (€20 per vCPU-month, €10 per GiB-month) with a hard monthly ceiling per replica, for quieter APIs, staging and review apps.
Normal app bandwidth is included. App builds cost €0.50/hour by running seconds. If the balance runs out, apps pause; no bill arrives afterwards.
The difference in monthly costs shows most on a backend that is busy all day. Example costs for a service with 1 vCPU and 2 GB of memory, serving traffic around the clock, in Frankfurt:
| Workload | Vercel Pro fra1 | Watasu |
|---|---|---|
| CPU busy 20% of the time | about $49/month | €15/month (Standard 1×) |
| CPU busy 100% of the time | about $156/month | €15/month (Standard 1×) |
| Same service, 2 replicas in separate zones | about 2× the above | €30/month |
| Team of 5 developers | +$80/month in seats | €0 |
| 5 TB of traffic per month | +$20/month CDN tier | €0 |
Vercel figures use Frankfurt rates of $0.184 per Active CPU hour and $0.0152 per GB-hour of memory, over a 730-hour month, with the $20 credit applied once and invocations not included. A project that sits idle almost all the time can cost less on Vercel, where the included credit covers it.
Key differences
Infrastructure: long-running processes vs serverless functions
On Vercel, your server code runs as functions. An instance starts when requests arrive, handles them, and is paused when traffic stops. Functions on the Pro plan run for 300 seconds by default and up to 800 seconds, with 1,800 seconds in beta. They get up to 2 vCPU and 4 GB of memory, and request and response bodies are capped at 4.5 MB. This works well for short API routes and server rendering. It gets harder for anything that has to keep running: queue consumers, schedulers, streaming agents, long uploads or persistent connections.
On Watasu, your app is a set of processes defined in a Procfile, and each one runs continuously in its own container:
webfor public HTTP and WebSockets, with managed TLSworkerfor queues, consumers and schedulers, not reachable from the internet*-tcpfor private TCP and gRPC services, reachable only by apps you allow*-rtcfor WebRTC and voice, with a dedicated TURN gateway and public UDPreleasefor migrations that run before each new version goes live
Each process gets its own size, from a small Hobby pod up to 16 vCPU and 30 GiB of RAM, and its own replica count. Once a process has two or more replicas, Watasu spreads them across separate zones at no extra charge.
Where your data lives
Vercel Inc. is a U.S. company. Vercel lists Washington, D.C. (iad1) as the default execution region for new projects. You can configure functions to run in an EU region such as Frankfurt, Paris or Dublin. Marketplace database locations depend on the provider you choose.
The U.S. CLOUD Act can apply even when data is hosted in the EU. Under 18 U.S.C. § 2713, valid U.S. legal process can require a covered provider to disclose data in its possession, custody or control, wherever that data is stored.
Watasu LTD is registered in England and Wales. Watasu hosts apps and add-ons in Germany, and customer workload data, including app images, databases, backups, logs, metrics, traces and object storage, stays in the EU.
Databases and backend services
Vercel no longer runs its own Postgres or key-value store. Databases come from Marketplace partners such as Neon, Supabase, AWS Aurora, Prisma and Upstash. Billing goes through Vercel, but each service is a separate provider with its own console, plans, region and terms.
Watasu runs its data services itself, next to your apps, under one bill:
| Service | What it's for | From |
|---|---|---|
| PostgreSQL | Application data; local NVMe, followers and high-availability standby on higher tiers | €5/month |
| Valkey | Redis-compatible cache, sessions, queues; HA with automatic failover on Premium | €1/month |
| ClickHouse | Analytics and event data, replicated across two zones on Standard | €35/month |
| Qdrant | Vector and hybrid search for RAG | €25/month |
| Redpanda | Kafka-compatible event streaming | €30/month |
| Object Storage | S3-compatible buckets with bundled egress | €2/month |
| Grafana, metrics, traces | Dashboards, alerts, OpenTelemetry tracing | Free tier |
Attach a service with one command and Watasu injects the connection variables into your app. Scheduled backups and a restore-and-promote workflow are built in on production tiers.
Real-time, voice and long-running work
Vercel added WebSocket support to Functions as a public beta. Each connection closes when the function reaches its maximum duration, so clients need reconnect logic, and shared state has to live in an external store because connections are pinned to individual instances. Always-on workers are available as Vercel Daemons, also in public beta, without inbound traffic, preview deployments or autoscaling. There is no UDP, so WebRTC media can't run on Vercel.
On Watasu these are ordinary processes. Web processes hold WebSocket connections, worker processes pull jobs for as long as there are jobs, and *-rtc processes get a TURN gateway, a public UDP port and a stable hostname for each replica. You bring your SFU, such as LiveKit Server, mediasoup or Pion. This makes Watasu a natural home for voice agents that need low latency for European users.
AI sandboxes
Both platforms offer isolated sandboxes for running AI-generated or untrusted code. Vercel Sandbox bills Active CPU, so you pay for CPU only while code is computing, plus provisioned memory, and its sandboxes run in US regions and Paris.
Watasu sandboxes run in the EU, next to the rest of your stack. You get reusable templates, SSH access, file and command APIs, pause and resume, disk checkpoints and outbound network policies. Runtime is billed per second for the vCPU and memory you request: a 2 vCPU, 1 GiB sandbox costs about €0.16 per hour. The Basic plan is free, with runs up to 1 hour and 20 concurrent sandboxes; Pro (€150/month) raises that to 24 hours and 100.
Developer workflow
On Vercel, each app is a project, and databases and other services are installed from the Marketplace and managed in each provider's dashboard.
On Watasu, one app holds all of its processes and attached services, and a team can share services across apps. Pipelines move the same tested image from staging to production without a rebuild. Access is set per app or pipeline with four roles: Viewer, Deployer, Operator and Maintainer. Every app gets its own Grafana workspace, and dashboards can live in your repository and sync on each release. You can also connect Claude or another AI assistant over MCP to inspect apps, query logs and manage resources.
When Vercel is the better fit
Vercel is a strong choice when your product is mostly a frontend:
- A marketing site or content-heavy Next.js app whose visitors are spread across every continent and who benefit from edge caching close to them.
- Short API routes with spiky, mostly idle traffic that stays inside the $20 monthly credit.
- Teams that rely on Vercel-specific products such as v0 or the AI Gateway.
If your product has a real backend (APIs that do work, queues, scheduled jobs, a database, real-time features or AI agents), and your users or customers are in Europe, Watasu runs the whole thing on one platform.
Migrate from Vercel to Watasu
Most Vercel projects move over in an afternoon. Sign up at dash.watasu.io and top up your balance to get started; if you'd like a hand, we help with the migration and give starting credits.
- Create the app and connect GitHub. Create an app in the dashboard or with the CLI, then connect your repository for automatic deploys, or push directly with
git push watasu main. - Define your processes. Add a Procfile at the root of the repository. Your Vercel cron jobs become a scheduler in a worker process.
web: npm run start worker: node worker.js release: npx prisma migrate deploy - Add your databases. Attach PostgreSQL, Valkey or Object Storage with one command, for example
watasu addons:create postgresql --app my-app. Watasu injects the connection variables. Move your data from Neon or Supabase withpg_dumpandpg_restore, and replace Vercel Blob with Object Storage. - Copy your environment variables. Pull them from Vercel with
vercel env pull, then set them as config vars on the Watasu app. - Point your domain. Add your custom domain to the app, update the DNS record, and Watasu issues and renews the certificate.
- Turn on review apps and pipelines. Connect a staging branch, enable review apps for pull requests, and promote tested builds to production without rebuilding.
Moving a larger workload, or want to see your stack running on Watasu before you switch? Write to info@watasu.io and we'll set it up with you.
Other comparisons
Frequently asked questions
Can I run Next.js on Watasu?
Yes. A Next.js app runs on Watasu as a regular web process with next start, next to its API, workers and database. Features such as server rendering, API routes, middleware and image optimisation run in the Node.js server itself.
Does Watasu have cold starts?
No. Processes stay running, so the first request after a quiet period is as fast as any other. Dynamic pods stay online too and bill only measured usage, from €0.30 per replica per month.
Is Watasu cheaper than Vercel?
For services that serve traffic all day, usually yes: a fixed €15 per month for 1 vCPU and 2 GiB, with bandwidth and team seats included; app builds cost €0.50/hour by running seconds. For a project that sits idle almost all the time, Vercel's included credit can cost less.
Is my data really in the EU?
Yes. Apps, databases, backups, logs, metrics, traces and object storage all run in data centers in Germany, and Watasu is not a US company.
Can I bring my own Docker image?
Yes. Add a Dockerfile at the root of your repository and Watasu builds it. Without one, Cloud Native Buildpacks detect your stack and build the image for you.
Put your next app on Watasu
Deploy from Git, keep your data in Germany, pay one prepaid balance in euros. Starting credits for teams moving from Vercel.