The right replacement for a serverless Postgres database depends on one decision: do you need a database on its own, or a whole backend with sign-in, storage and APIs wrapped around it? Answer that first. Then compare branching, scale to zero and cold starts on your own schema, across managed Postgres, distributed SQL and single-job specialists.
Thinking of moving off Neon? You're not alone, because the business behind it has changed shape, and that's prompted plenty of developers to look at where else their data could live. Here are eight alternatives, in no particular order, sorted by what you're really buying rather than by how big the company is.
What's changed behind the serverless Postgres brand
Neon's own homepage now describes the product as running on Lakebase Postgres, with a gateway powered by Databricks. In plain terms, the database you signed up for now sits inside a much larger data company.
That doesn't mean anything breaks tomorrow. Your connection strings still work and your data's still there. But the roadmap, the support team and the contract terms may now be shaped by a business with very different priorities from a small developer-first startup. That's reason enough to know where you'd go, even if you decide to stay.
It's also a good moment to ask whether serverless Postgres was the right fit in the first place. Some teams chose it for branching. Others chose it because it scaled to zero and kept side projects tidy. A few picked it as the database half of a backend they then had to assemble themselves. Each of those reasons points to a different replacement.
Database only, or a whole backend?
The tools below split into two camps, and mixing them up wastes weeks. One camp gives you a database and nothing else. The other wraps the database in sign-in, file storage, APIs and functions, so a small team can ship an app without stitching five services together.
- If you already run your own auth and storage, you want a database. Look at the managed, distributed and specialist options.
- If you were planning to add auth, storage and an API layer anyway, a backend platform saves you building it.
- If your data has an unusual shape, such as one database per customer or a stream of sensor readings, start with the specialists.
Write your answer down before you open a single sign-up form. It'll keep the shortlist short.
Eight database options side by side
| Tool | Kind of database | Good for |
|---|---|---|
| Aiven for PostgreSQL | Managed Postgres | Managed Postgres with extensions, forking and pooling |
| Crunchy Data | Enterprise Postgres | Running on cloud, Kubernetes or enterprise with support |
| Nile | Multi-tenant Postgres | B2B SaaS apps with many customer tenants |
| Supabase | Postgres backend platform | A full backend around a Postgres database |
| TiDB Cloud | MySQL-compatible serverless SQL | Distributed SQL that scales past a single node |
| Tiger Data (Timescale) | Time-series Postgres | Sensor, machine and other time-series data in Postgres |
| Turso | SQLite-based databases | A separate database for each user, agent or tenant |
| Xata | Branch-native Postgres | Instant branching and scale to zero on Postgres |
Four kinds of replacement
A full backend platform. Supabase is the one widely known name here. It puts a Postgres database at the centre and adds authentication, instant APIs, realtime, functions, storage and vector embeddings around it. If you were using serverless Postgres as one piece of a backend, this is the closest thing to getting the other pieces in the same box.
Distributed SQL. TiDB Cloud comes from PingCAP and is built for auto-scaling, distributed workloads. The thing to notice is that it's MySQL-compatible first, and the vendor describes Postgres compatibility as a public preview. If your app leans on Postgres features, test that carefully before you plan a move.
Managed Postgres from infrastructure vendors. Aiven for PostgreSQL offers managed Postgres with extensions, database forking and connection pooling, and forking is the nearest thing here to the branching you may be used to. Crunchy Data runs managed Postgres for cloud, Kubernetes or enterprise setups and sells commercial support alongside it. Both are about running standard Postgres well, rather than reinventing it.
Tools built for one job. Nile re-engineers Postgres for multi-tenant B2B apps, so each customer's data is handled as its own tenant. Turso takes a different route entirely, built on SQLite and pitched at giving every user, agent or tenant its own database. Xata keeps vanilla PostgreSQL but makes instant branching and scale to zero the headline features. Tiger Data (Timescale) is Postgres from the TimescaleDB creators, aimed at sensor and machine data with automatic partitioning.
Notice that only one group offers a whole backend. If that's what you need, the other three groups will leave you building sign-in and storage yourself.
One line on each database
Listed A to Z.
Aiven for PostgreSQL
Best for: Teams wanting managed PostgreSQL with extensions, database forking and connection pooling; the vendor also mentions a free tier for learning the platform.
Crunchy Data
Best for: Organisations running Postgres across cloud, Kubernetes or enterprise setups who want commercial support, with the vendor stressing open-source Postgres and no proprietary lock-in.
Nile
Best for: Builders of multi-tenant B2B apps who want Postgres re-engineered around tenants, including customer-specific vector embeddings through pgvector, as the vendor describes it.
Supabase
Best for: Product teams who want a Postgres database plus authentication, instant APIs, realtime, functions, storage and vector embeddings in one platform.
TiDB Cloud
Best for: Developers comfortable with MySQL who need auto-scaling distributed SQL from PingCAP; the vendor lists its Postgres compatibility as a public preview.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
Tiger Data (Timescale)
Best for: Engineers storing sensor and machine data who want Postgres from the TimescaleDB creators, with automatic partitioning for live and historical data, per the vendor.
Turso
Best for: Apps that give every user, agent or tenant its own database, built on SQLite with an open-source SQLite-compatible engine.
Xata
Best for: Anyone who wants vanilla PostgreSQL with instant branching built in and scale to zero when a database sits idle, in the vendor's words.
My advice before you move a single table
Don't migrate because of a headline. A change of owner is a reason to make a plan, not a reason to panic. Work out which feature you'd miss most, whether that's branching for preview environments, scale to zero for quiet projects or the simplicity of one connection string, and judge every alternative on that feature first.
If you're a small team without a backend yet, a platform that bundles auth and storage will save you more time than any database tweak. If you've already got those pieces, stay with plain managed Postgres and keep your options open, because standard Postgres moves between providers far more easily than anything proprietary.
Be wary of switching database engines just to switch vendors. Moving from Postgres to a MySQL-compatible or SQLite-based system can be the right call, but it's a rewrite of queries and tooling, not a migration. For more neighbours in this space, see my list of PlanetScale alternatives, the head-to-head of the two best-known serverless Postgres names and my guide to app hosting alternatives.
Three tests to run on your own schema
- Branching: create a branch from a copy of your real database, run a migration on it and time how long the branch takes to appear and to delete.
- Scale to zero and cold starts: leave a test database idle overnight, then hit it with your app's first real query in the morning and see how long that first answer takes.
- Your extensions and queries: load your actual schema, including every extension you rely on, and run your slowest production queries. A demo database won't show you where things break.
Run those three tests on your two shortlisted tools and you'll know more than any comparison chart can tell you.
Do you make a Postgres or serverless database platform? Tool makers can ask to be considered for a featured place on this list.
Want your tool on this list? Featured spots and reviews here are paid partnerships. Here's how to partner with me.
Frequently asked questions
Is serverless Postgres different from managed Postgres?
Serverless Postgres usually separates storage from compute, so the database can pause when idle and wake when a query arrives. Managed Postgres runs a standard server that a provider looks after for you, and it's typically always on.
What's database branching?
Branching makes a quick copy of a database, schema and data, so you can test a migration or build a preview environment without touching production. Some platforms build it in, while others offer forking, which is similar.
Do I need a backend platform or just a database?
If you already run your own sign-in, file storage and API layer, a database is enough. If you'd have to build those pieces, a backend platform with them included will usually get a small team shipping sooner.
Can a MySQL-compatible database replace Postgres?
It can, but it's a bigger job than switching Postgres providers. Queries, extensions and tooling may all need changing, so check what Postgres support the vendor offers and test your own queries first.
Is SQLite a sensible choice for a SaaS app?
For some apps, yes, especially where each user or customer gets their own small database. For apps that need one large shared dataset with heavy concurrent writes, a server database is usually the safer fit.
How do I test cold starts fairly?
Leave the database idle for as long as your quietest real period, then send the first query your app would send. Repeat it a few times on different days, because a single test can mislead you.
